Here are some of the the major AI trends shaping 2026 — based on current expert forecasts, industry reports, and recent developments in technology. The material is analyzed using AI tools and final version hand-edited to this blog text:
1. Generative AI Continues to Mature
Generative AI (text, image, video, code) will become more advanced and mainstream, with notable growth in:
* Generative video creation
* Gaming and entertainment content generation
* Advanced synthetic data for simulations and analytics
This trend will bring new creative possibilities — and intensify debates around authenticity and copyright.
2. AI Agents Move From Tools to Autonomous Workers
Rather than just answering questions or generating content, AI systems will increasingly act autonomously, performing complex, multi-step workflows and interacting with apps and processes on behalf of users — a shift sometimes called agentic AI. These agents will become part of enterprise operations, not just assistant features.
3. Smaller, Efficient & Domain-Specific Models
Instead of “bigger is always better,” specialized AI models tailored to specific industries (healthcare, finance, legal, telecom, manufacturing) will start to dominate in many enterprise applications. These models are more accurate, legally compliant, and cost-efficient than general models.
4. AI Embedded Everywhere
AI won’t be an add-on feature — it will be built into everyday software and devices:
* Office apps with intelligent drafting, summarization, and task insights
* Operating systems with native AI
* Edge devices processing AI tasks locally
This makes AI pervasive in both work and consumer contexts.
5. AI Infrastructure Evolves: Inference & Efficiency Focus
More investment is going into inference infrastructure — the real-time decision-making step where models run in production — thereby optimizing costs, latency, and scalability. Enterprises are also consolidating AI stacks for better governance and compliance.
6. AI in Healthcare, Research, and Sustainability
AI is spreading beyond diagnostics into treatment planning, global health access, environmental modeling, and scientific discovery. These applications could help address personnel shortages and speed up research breakthroughs.
7. Security, Ethics & Governance Become Critical
With AI handling more sensitive tasks, organizations will prioritize:
* Ethical use frameworks
* Governance policies
* AI risk management
This trend reflects broader concerns about trust, compliance, and responsible deployment.
8. Multimodal AI Goes Mainstream
AI systems that understand and generate across text, images, audio, and video will grow rapidly, enabling richer interactions and more powerful applications in search, creative work, and interfaces.
9. On-Device and Edge AI Growth
10. New Roles: AI Manager & Human-Agent Collaboration
Instead of replacing humans, AI will shift job roles:
* People will manage, supervise, and orchestrate AI agents
* Human expertise will focus on strategy, oversight, and creative judgment
This human-in-the-loop model becomes the norm.
Sources:
[1]: https://www.brilworks.com/blog/ai-trends-2026/?utm_source=chatgpt.com “7 AI Trends to Look for in 2026″
[2]: https://www.forbes.com/sites/bernardmarr/2025/10/13/10-generative-ai-trends-in-2026-that-will-transform-work-and-life/?utm_source=chatgpt.com “10 Generative AI Trends In 2026 That Will Transform Work And Life”
[3]: https://millipixels.com/blog/ai-trends-2026?utm_source=chatgpt.com “AI Trends 2026: The Key Enterprise Shifts You Must Know | Millipixels”
[4]: https://www.digitalregenesys.com/blog/top-10-ai-trends-for-2026?utm_source=chatgpt.com “Digital Regenesys | Top 10 AI Trends for 2026″
[5]: https://www.n-ix.com/ai-trends/?utm_source=chatgpt.com “7 AI trends to watch in 2026 – N-iX”
[6]: https://news.microsoft.com/source/asia/2025/12/11/microsoft-unveils-7-ai-trends-for-2026/?utm_source=chatgpt.com “Microsoft unveils 7 AI trends for 2026 – Source Asia”
[7]: https://www.risingtrends.co/blog/generative-ai-trends-2026?utm_source=chatgpt.com “7 Generative AI Trends to Watch In 2026″
[8]: https://www.fool.com/investing/2025/12/24/artificial-intelligence-ai-trends-to-watch-in-2026/?utm_source=chatgpt.com “3 Artificial Intelligence (AI) Trends to Watch in 2026 and How to Invest in Them | The Motley Fool”
[9]: https://www.reddit.com//r/AI_Agents/comments/1q3ka8o/i_read_google_clouds_ai_agent_trends_2026_report/?utm_source=chatgpt.com “I read Google Cloud’s “AI Agent Trends 2026” report, here are 10 takeaways that actually matter”
2,513 Comments
Tomi Engdahl says:
https://venturebeat.com/orchestration/researchers-introduce-self-harness-a-framework-that-lets-ai-agents-rewrite-their-own-rules-boosting-performance-up-to-60
Tomi Engdahl says:
https://www.cyberday.ai/guides/frameworks-in-finland?utm_term=&utm_campaign=nl_Demand+Gen+_guide_+2025&utm_source=adwords&utm_medium=ppc&hsa_acc=7610695024&hsa_cam=23787822604&hsa_grp=198583421729&hsa_ad=806399046405&hsa_src=&hsa_tgt=aud-2439304165333&hsa_kw=&hsa_mt=&hsa_net=adwords&hsa_ver=3&gad_source=1&gad_campaignid=23787822604&gbraid=0AAAAADnQLxNQPfSqw_zgzArjaim_qMesw&gclid=CjwKCAjwxb7RBhA5EiwAQ-AAdHYlhAIzIIa4GGWRS_s6AInUFBfoCWmrv2YkaJ6zy0G1vqZXQ7dLRhoCN3UQAvD_BwE
Tomi Engdahl says:
https://simonwillison.net/2026/jun/17/glm-52/
Tomi Engdahl says:
https://linas.substack.com/p/glm-52-local-ai-guide
Tomi Engdahl says:
https://www.oreilly.com/radar/linear-thinking-nonlinear-costs/
Tomi Engdahl says:
https://github.com/GoogleCloudPlatform/knowledge-catalog/blob/main/okf/SPEC.md
Tomi Engdahl says:
https://www.howtogeek.com/google-is-not-the-best-search-engine-anymore/
Tomi Engdahl says:
Your team picks a new AI vendor every week. Are they safe?
Legitima scans any AI vendor in 30 seconds. Get risk signals, compliance gaps, and required actions — before procurement signs.
https://legitima.ai/
Tomi Engdahl says:
https://etn.fi/index.php/13-news/19093-check-point-tuo-openai-n-kybermallit-tietoturvatuotteisiinsa
Tomi Engdahl says:
Amerikka veti tappokytkimestä – Nyt paljastui syy
Yhdysvalloissa säikähdettiin vihamielisten valtioiden mahdollisuuksia käyttää Anthropicin tekoälyä. Tapaus herättää monia kysymyksiä.
https://www.iltalehti.fi/digiuutiset/a/3b74422d-2583-433a-a909-3faed20ae6c5
Tekoäly-yhtiö Anthropic julkaisi Fable 5 -kielimallin, jolla ei pitäisi voida löytää vakavia tietoturvahaavoittuvuuksia. Viranomaisten pelkona on, että turvarajoitukset voidaan kiertää.
Tekoäly-yhtiö Anthropic teki kovan päätöksen sulkea kehittyneimmät tekoälymallinsa kaikilta käyttäjiltä, kun Yhdysvaltain hallinto vaati sitä estämään mallien käytön muilta kuin Yhdysvaltojen asukkailta.
Taustalla on hallinnon pelko siitä, että vihamieliset valtiot voisivat käyttää tekoälyä sotilastiedustelun edistämiseen, uutisoi Reuters.
Yhdysvaltain hallinnon edustajat ja Anthropicin työntekijät sekä johto kävivät neuvotteluja tilanteesta viikonlopun ja maanantain aikana. Anthropic pyrkii palauttamaan tekoälymallit käyttäjien saataville. Yhdysvaltain hallinto haluaa takeita siitä, ettei uusia malleja voida käyttää Yhdysvaltojen vahingoittamiseen.
Fable 5 -tekoäly on varustettu turvarajoituksilla, jotka estävät sen käytön tietoturvahaavoittuvuuksien löytämiseen. Viranomaisten huolena on, että nämä rajoitukset voitaisiin kiertää.
https://www.reuters.com/technology/anthropic-us-officials-meeting-monday-resolve-dispute-over-export-curbs-2026-06-15/
Tomi Engdahl says:
https://www.nytimes.com/2026/06/23/us/politics/nsa-lost-access-anthropic-tool.html?unlocked_article_code=1.sVA.7kXO.45D5zt8PSNGg&smid=url-share
#2600net #irc #secnews via Dave Schroeder
Tomi Engdahl says:
AI in games may be hurting more than helping, with research linking disclosed AI use to major drops in sales and reviews For big studios, the AI stigma is becoming impossible to ignore!!
Read more: https://bit.ly/43RmI3O
#GamingNews #AI #GameDevelopment
AI Use In Games Leads To 40-60% Drop In Sales, Study Reveals
https://tech4gamers.com/ai-use-games-drop-sales/?fbclid=IwdGRjcASo45hjbGNrBKjjb2V4dG4DYWVtAjExAHNydGMGYXBwX2lkDDM1MDY4NTUzMTcyOAABHkIq0C2O3F5RsTpgadzJheDZOwT4QK6IbQw99DLQOGTePve1sFSJYSgsD0y3_aem_Y3kGaQnPRCpwpLw3cq0-ww
AI Stigma Severely Impacts A Game’s Performance After Launch!
Story Highlight
AI use in games can lead to a 40-60% drop in sales for established developers.
Titles that disclose AI use received 53% fewer reviews than non-AI games under similar conditions.
The research shows AI stigma is real and hits large-scale, established studios the hardest.
AI use in gaming has grown over the years, with generative AI now often being used to create assets, streamline development, or even cut corners that harm developers in one way or another. Therefore, there’s been a lot of controversy around the technology.
A study shows that disclosing AI use on Steam slashes a game’s reviews by anywhere around 53%, which translates to a concerning 40-60% sales drop for established studios. Without AI, these games would have enjoyed a 20% to 60% sales boost instead via marketing.
Why it matters: Research shows that embracing AI use in games becomes a liability for established studios, even as the industry moves to accept it. In other words, games relying on AI would have performed much better without it.
Game Oracle’s research analyzed nearly 10K Steam releases that launched last year before November and found that 21% disclosed AI use.
Nearly 20% of titles using AI didn’t receive any reviews as opposed to 15% for non-AI ones. Similarly, another set of titles that used AI only averaged around 4 reviews in their first month after launch, compared to 7 reviews for those developed without it.
For titles that received at least 100 reviews, those without AI hit higher average scores (88.3%) than games that used the technology (84.6%).
The study also found that inexperienced developers are less likely to suffer setbacks with AI use since their projects would struggle without it anyway. So, large AAA studios that experiment with AI in game development experience the biggest AI stigma in the industry.
if “good” studios are using AI — then AI use is catastrophic (-40% to -60% drop in sales). This is evidenced by the dark blue cells at the top of our heatmap.
-Game Oracle.
Tomi Engdahl says:
Is it really an “arms race” when each side differs by days? https://www.straitstimes.com/world/chinas-360-says-it-has-developed-tools-to-match-anthropics-mythos
#2600net #irc #secnews #isc360
Tomi Engdahl says:
Oracle spent $55.7B on AI. 21,000 jobs disappeared.
Staff fell from 162,000 to 141,000 in one year.
Its official SEC filing names AI as the cause.
Whole database teams were handed to AI agents.
The AI tools you use now replace real roles.
Most companies hide these cuts behind softer words.
Read more on TNW: https://thenextweb.com/news/oracle-21000-layoffs-ai-data-centres
Tomi Engdahl says:
From Phoenix to Melbourne, a global coalition of city leaders says the AI data centre boom is draining their power, water and land — and they are done waiting.
https://l.euronews.com/J24I
Tomi Engdahl says:
A fintech company said its employee burned through $80,000 in tokens making a ‘brainrot shooter game’ : https://mrf.lu/nmBS
Tomi Engdahl says:
Fully autonomous drones have killed human soldiers for the first time
A senior figure in the Ukrainian defence industry told New Scientist that a test took place two years ago involving fully autonomous drones set to destroy anything in a given area, with confirmed casualties
https://www.newscientist.com/article/2529849-fully-autonomous-drones-have-killed-human-soldiers-for-the-first-time/?fbclid=IwdGRjcASpHydjbGNrBKkfIGV4dG4DYWVtAjExAHNydGMGYXBwX2lkDDM1MDY4NTUzMTcyOAABHsB9-KIYCzt84_P2WQxqUfEM5Q0C28RQXNi7IaQ5oijEp2LXttD2FuXH4lbd_aem_TAcI-YHCl-xNqmRhXGt5gA&utm_id=97758_v0_s00_e227_tv4_tp1_a1dennhavpqa97
Tomi Engdahl says:
“We just launch it and we know everything will be dead — everything that will be found there in this particular area will be dead.” https://trib.al/wLZVjJt
Tomi Engdahl says:
Full story: https://www.headphonesty.com/2025/09/audiophile-ruins-system-bad-ai-advice/?utm_source=fb&utm_campaign=comment
Tomi Engdahl says:
Elon Musk’s xAI is reportedly leaning into explicit content generation as a core driver of its Grok chatbot traffic and adult content now accounts for the majority of the platform’s activity, according to a new report.
Read more: https://www.forbes.com/sites/maryroeloffs/2026/06/24/groks-traffic-is-mostly-driven-by-adult-content-report-says/?utm_source=ForbesMainIG&utm_medium=social&utm_campaign=ForbesMainIG
Tomi Engdahl says:
“Osa Helsingin Kauppatorilla myytävistä matkamuistoista on kuvitettu tekoälyllä.” – Toivottavasti kenellekään ei tule yllätyksenä, että suurin osa matkamuistoista on halpaa kiinalaista rihkamaa ja kyynistä rahastusta ja että myös onnittelu- ja joulukorteista suuri osa on tuotettu tekoälyllä, siis myös niistä brändillisistä joita isot kaupat myyvät.
https://www.hs.fi/helsinki/art-2000012067211.html?fbclid=IwdGRjcASpvpVleHRuA2FlbQIxMQBzcnRjBmFwcF9pZAwzNTA2ODU1MzE3MjgAAR4LY9KZsqvCyDbceROH8jJS8-GRQG9Qo_cPcRt45Ldv6zQqO-OijpAiF4xwzA_aem_V_6Es9c5YvmSPtQCbkUpGQ
Kauppatorilla myydään Helsinkiä, jota ei ole olemassa
Osa Helsingin Kauppatorilla myytävistä matkamuistoista on kuvitettu tekoälyllä.
Helsinki-kuvaan sekoittuu myös poroja, revontulia ja viikinkiviittauksia.
Tutkijan mukaan matkamuisto voi muodostaa vastaanottajalleen ainoan mielikuvan matkakohteesta.
Helsingin tuomiokirkko kohoaa jääkaappimagneeteissa ja postikorteissa yhä uudelleen, mutta sen väri, kupolit ja korinttilaisten pylväiden lukumäärä vaihtelevat.
Toisinaan Helsingin siluettiin on ilmestynyt Tallinnan vanhaa kaupunkia muistuttava mustahuippuinen kirkontorni. Osassa matkamuistoista Uspenskin katedraalin on korvannut ilmeisesti Keski-Porin kirkko.
Kauppatorilla Helsinki-mielikuva on myynnissä muutamalla eurolla postikorttien, magneettien, lippisten, kangaskassien ja kenkälusikoiden muodossa.
Kojussa tuotteita myy monen muun tavoin Scandifors Oy:n Natalia Prozorova. Kun häneltä kysyy Tallinnaa muistuttavasta kirkontornista, Prozorova kertoo tuotteiden olevan massatuotettuja mutta yrityksen itse suunnittelemia.
Joidenkin tuotteiden suunnittelussa on Prozorovan mukaan käytetty tekoälyä.
Tavoitteena ei hänen mukaansa aina ole realismi vaan helsinkiläisen maiseman luominen. Turistit ostavat etenkin keskeisiä nähtävyyksiä esittäviä matkamuistoja, mutta myös Lappi-kuvastolla poroineen ja revontulineen on Helsingissä paikkansa.
Paikalliselle väärä kirkontorni voi näyttäytyä huvittavana särönä ja lappilais-Helsinki erikoisena vaihtoehtotodellisuutena. Matkailututkimuksen näkökulmasta se on kuitenkin osa kaupungista maailmalle lähtevää kuvaa
Lapin yliopiston matkailututkija Monika Lüthje arvioi, että turistin mukana kulkeutuva matkamuisto voi olla ainoa mielikuva, jonka kotona olevat saavat matkakohteesta. Jos paikallista kontekstia ilmennetään tekoälyn avulla tai puutteellisin tiedoin, lopputulos voi hänen mukaansa olla ”aika hurja”.
Tomi Engdahl says:
https://etn.fi/index.php/13-news/19103-ai-palvelimissa-kellotus-nousee-uuteen-rooliin
Tomi Engdahl says:
https://etn.fi/index.php/13-news/19102-openai-suunnitteli-oman-llm-kiihdyttimen-broadcomin-kanssa
OpenAI on ottanut uuden askeleen kohti täyttä tekoälypinoa. Broadcomin kanssa kehitetty Jalapeno ei ole yleiskäyttöinen prosessori, vaan suurten kielimallien inferenssiin optimoitu ASIC-kiihdytin, jolla OpenAI hakee parempaa energiatehokkuutta, pienempää viivettä ja vähemmän riippuvuutta ulkopuolisista tekoälykiihdyttimistä.
OpenAI kutsuu Jalapenoa ensimmäiseksi Intelligence Processor -piirikseen. Kyse ei siis ole yleiskäyttöisestä prosessorista eikä perinteisestä grafiikkapiiristä, vaan asiakaskohtaisesta ASIC-ratkaisusta, jonka arkkitehtuuri on optimoitu OpenAI omien mallien ja palveluiden tarpeisiin.
OpenAI mukaan Jalapenon suunnittelussa on hyödynnetty yhtiön omaa näkemystä siitä, miten ChatGPT, Codexin, API-palveluiden ja tulevien agenttituotteiden kuormat käyttäytyvät käytännössä. Optimoinnin kohteina ovat olleet muun muassa mallien käyttämät kernelit, muistinsiirto, verkotus ja palvelinkuormien ajo datakeskuksissa. Tavoitteena on saada todellinen käyttöaste lähemmäs sirun teoreettista huippusuorituskykyä.
Tomi Engdahl says:
Artificial Intelligence
When Information Becomes the Attack Surface – Understanding AI Agent Traps
From hidden content injections to cognitive state poisoning, attackers are turning trusted data sources into traps for autonomous AI.
https://www.securityweek.com/when-information-becomes-the-attack-surface-understanding-ai-agent-traps/
AI agents go beyond answering questions. They can autonomously browse websites, read emails, search company files, query software tools, and more. AI models producing incorrect answers is hardly a threat, until agents encounter information that’s maliciously designed to influence what it sees, believes, remembers, or executes.
An agent leverages webpages, document stores, wikis, images, emails, or tools to produce intended outputs. But what happens when these sources mask malicious instructions? These trap AI agents into making a wrong interpretation or taking unintended action. Scientists from Google DeepMind categorized these “traps” into six categories, including content injection, semantic manipulation, cognitive state, behavioral control, systemic, and human-in-the-loop traps. The last two are more theoretical and expected to become more relevant as AI agent use grows. It helps to understand these traps to determine the necessary mitigations.
Content Injection: When Instructions Hide in Plain Sight
Content injections exploit the difference between what a human sees and what an agent parses, as well as the system’s difficulty in keeping trusted instructions separate from untrusted external data.
A webpage might appear harmless, but its underlying code, metadata, hidden text, or image can contain malicious instructions for an AI system. An AI model accepts attacker-controlled data from an external source, such as a website or file. If this system fails to distinguish between data and instructions, the model may start processing instructions within that content. The objective behind such injection of malicious content is to alter the AI’s response, disclose sensitive information or enable an unauthorized action. In NIST evaluations of agent hijacking, malicious instructions succeeded across five tested injection tasks, on average, 57% of the time.
Semantic Manipulation: Shapeshifting the Information
Semantic manipulation need not explicitly tell the agent what to do; it feeds repetition, emotional language, selective context, a false sense of authority, and coordinated claims to the agent to skew context and guide the agent towards the ‘attacker preferred’ conclusion.
Imagine a scenario where you have tasked an agent to zero in on a supplier. It comes across search results that repeatedly extol the virtues of a specific supplier, describe a specific company as the gold standard, highlight its strengths and amplify doubts about competitors. This increases the chances of the agent recommending this supplier. Conventional signature-based security tools may not flag anything malicious, as the attacks leverage ‘reasoning’ to influence rather than rely on malicious code.
AI agents go beyond answering questions. They can autonomously browse websites, read emails, search company files, query software tools, and more. AI models producing incorrect answers is hardly a threat, until agents encounter information that’s maliciously designed to influence what it sees, believes, remembers, or executes.
An agent leverages webpages, document stores, wikis, images, emails, or tools to produce intended outputs. But what happens when these sources mask malicious instructions? These trap AI agents into making a wrong interpretation or taking unintended action. Scientists from Google DeepMind categorized these “traps” into six categories, including content injection, semantic manipulation, cognitive state, behavioral control, systemic, and human-in-the-loop traps. The last two are more theoretical and expected to become more relevant as AI agent use grows. It helps to understand these traps to determine the necessary mitigations.
Content Injection: When Instructions Hide in Plain Sight
Content injections exploit the difference between what a human sees and what an agent parses, as well as the system’s difficulty in keeping trusted instructions separate from untrusted external data.
A webpage might appear harmless, but its underlying code, metadata, hidden text, or image can contain malicious instructions for an AI system. An AI model accepts attacker-controlled data from an external source, such as a website or file. If this system fails to distinguish between data and instructions, the model may start processing instructions within that content. The objective behind such injection of malicious content is to alter the AI’s response, disclose sensitive information or enable an unauthorized action. In NIST evaluations of agent hijacking, malicious instructions succeeded across five tested injection tasks, on average, 57% of the time.
A support ticket with underlying malicious instructions can manipulate an AI agent into retrieving customer data from the CRM and sending it to an attacker-controlled address. If the agent has excessive permission, this exfiltration becomes all the easier.
Semantic Manipulation: Shapeshifting the Information
Semantic manipulation need not explicitly tell the agent what to do; it feeds repetition, emotional language, selective context, a false sense of authority, and coordinated claims to the agent to skew context and guide the agent towards the ‘attacker preferred’ conclusion.
Advertisement. Scroll to continue reading.
AI Risk Summit 2026 – Get Your Ticket
Imagine a scenario where you have tasked an agent to zero in on a supplier. It comes across search results that repeatedly extol the virtues of a specific supplier, describe a specific company as the gold standard, highlight its strengths and amplify doubts about competitors. This increases the chances of the agent recommending this supplier. Conventional signature-based security tools may not flag anything malicious, as the attacks leverage ‘reasoning’ to influence rather than rely on malicious code.
Here, manipulation of the surrounding information environment becomes the manipulation of the decision itself.
Cognitive State Traps: Poisoning Agent Knowledge
Some agent systems use retrieval databases, interaction histories, or persistent memory stores to maintain context and continuity across tasks. This creates an opportunity for poisoned information to influence later outputs or actions. E.g., a poisoned document in a shared repository that an agent refers to and trusts as evidence, or a manipulated exchange that becomes an agent’s memory, only to rear its head during future tasks.
Research presented at the USENIX conference found that, in controlled tests, inserting five specially crafted texts per target question caused a RAG system to produce the attacker’s chosen answer in about 90% of cases, even when its knowledge base contained millions of legitimate texts.
With information governance becoming an integral component of AI security, organizations must be aware of which sources agents retrieve information from, who can modify those sources, how claims can be verified, and whether stored memories can be reviewed or removed.
Behavioral Control: Turning Influence into Action
Behavioral control operates at the juncture where interpretation is translated into action. Malicious content may attempt to make the AI agent send data, approve a transaction, execute code, invoke another tool or trigger a myriad of other actions. Here, the extent of the consequence depends on the extent of the agent’s access. Grant the agent only the data access and tool permissions required for the specific task. This could be the difference between an agent delivering a misleading summary and the same agent reading confidential files and communicating this information externally, resulting in data loss.
The More Theoretical Frontier
Systemic traps and human-in-the-loop traps remain less developed, but they deserve attention. Systemic traps could induce many similar agents to behave in correlated ways, causing congestion, market disruption, or cascading failures. Human-in-the-loop traps could use a compromised agent to mislead the person expected to approve its actions.
Control for Agent Traps
A single control won’t alleviate the agent trap threat. A defensive framework must have aspects like source verification, content screening, memory governance, restricted permissions, isolated execution, monitoring, and an independent approval framework with a human in the loop for high-impact actions. Security must follow authority, and there should be clear lines of separation between the ability to interpret and the authority to act.
The future of agentic AI use will depend not only on what these agents can do but also on how they decide what to trust.
Tomi Engdahl says:
Yifan Yu / Nikkei Asia:
Qualcomm CEO Cristiano Amon says the company is designing data center chips specifically for Chinese customers that are in compliance with US export controls — NEW YORK — Qualcomm unveiled its data center chip lineup on Wednesday, becoming the latest chipmaker to enter the AI processor race …
https://asia.nikkei.com/business/technology/artificial-intelligence/qualcomm-to-design-china-specific-data-center-chip-in-line-with-us-export-curbs
Kif Leswing / CNBC:
Qualcomm unveils Dragonfly C1000, a more power-efficient data center CPU built for agentic AI, and says Meta will use it when Qualcomm starts production in 2028
Qualcomm stock pops 15% after chipmaker almost doubles projection for 2029 non-handset revenue
https://www.cnbc.com/2026/06/24/qualcomm-data-center-cpu-meta.html
Reuters:
Qualcomm expects $15B in data center chip sales by 2029, raises its non-handset chip revenue forecast to $40B by 2029, up from $22B; QCOM jumps 13%+ pre-market
https://www.reuters.com/business/retail-consumer/qualcomm-bets-ai-chips-break-smartphone-reliance-faces-crowded-race-2026-06-24/
Tomi Engdahl says:
Financial Times:
Sources: Meta has accelerated its plans to use LLMs to review content and ads, replacing ~50% of human review requests in 2026 and aiming for 90%+ by year-end — Facebook parent is accelerating plans to use large language models to review content and ads across its platforms.
https://www.ft.com/content/39251a31-4a9d-4870-b86c-dc6353d67fdd
Tomi Engdahl says:
Financial Times:
As China’s working-age population shrinks, consensus is growing that China must embed embodied AI robots into as many tasks as possible, as soon as possible — The country’s workforce is set to fall to 300mn by the end of the century. Beijing wants humanoids to narrow the labour gap.
https://www.ft.com/content/c8731833-10ca-4a12-bfe4-8ebb2584ec68?accessToken=zwAAAZ79MZqMkdPIcxgzEMpKEtO_5I67JYTsaA.MEYCIQCSSI7J1vAXqRtz4tQYz551einE7UTB8Wv8AU2QBen0GQIhALM30pyYZWhUnmpAkZ2eHsndxmCSNLUirqcU-QmlRjhT&sharetype=gift&token=b45976a5-02d6-46b2-af02-f02579ed31bd
Tomi Engdahl says:
Kif Leswing / CNBC:
Micron reports Q3 revenue up 346% YoY to $41.46B, above $35.84B est., gross margin above est., and forecasts Q4 revenue above est.; MU jumps 14%+ pre-market — Micron’s revenue more than quadrupled in the fiscal third quarter, the company said on Wednesday, as the memory maker continued …
https://www.cnbc.com/2026/06/24/micron-mu-earnings-report-q3-2026.html
Tomi Engdahl says:
Jacob Reid / Bloomberg:
Exponential View: global AI sales, excluding China, hit $25B in Q1, exceeding an estimated $21B in data center and chip depreciation costs; margins remain thin — Revenue from artificial intelligence has reached a tipping point, showing that the hundreds of billions of dollars tech companies …
https://www.bloomberg.com/news/articles/2026-06-25/ai-demand-begins-to-justify-massive-cost-of-data-center-buildout
Tomi Engdahl says:
Tina Li / Wall Street Journal:
Mirendil, founded by former Anthropic researchers and seeking to build self-improving AI for open-source developers, raised a $200M seed at a $1B valuation — Mirendil raises $200 million from Andreessen Horowitz and Kleiner Perkins to make AI that can do the job of an AI engineer
https://www.wsj.com/tech/ai/anthropic-veterans-startup-seeks-to-help-scientists-develop-their-own-ai-09e2f3e5?st=CWXZSf&reflink=desktopwebshare_permalink
Tomi Engdahl says:
Lydia DePillis / New York Times:
OpenAI, Anthropic, Amazon, Microsoft, and others launch Raise Us, a new non-profit led by ex-Commerce Secretary Gina Raimondo to help US workers adapt to AI — OpenAI, Anthropic, Amazon and Microsoft have signed on to an effort led by Gina Raimondo, a former commerce secretary.
https://www.nytimes.com/2026/06/25/business/economy/ai-work-force-training-job-losses.html
Tomi Engdahl says:
Bloomberg:
NAND flash maker Kioxia, Japan’s most valuable company since June 12, plans to offer US depositary shares in spring 2027 amid AI-driven demand for memory chips — Kioxia Holdings Corp. plans to offer US depositary shares in the spring of 2027 and a stock split to take advantage of runaway demand …
https://www.bloomberg.com/news/articles/2026-06-25/japan-s-kioxia-plans-to-offer-us-depositary-shares-next-spring
Tomi Engdahl says:
After reassigning 7,000 employees into AI initiatives, Meta is now offering some engineers a path to transfer elsewhere.
#Meta #Metaemployee #techjobs #AI
Meta forced thousands of engineers into AI training work. Now it’s giving some a way out. : https://mrf.lu/npFj
Tomi Engdahl says:
NVIDIA just told its grey market: good luck without us.
CEO Jensen Huang spoke at the company’s shareholder meeting.
He said smuggled chips cannot build working AI data centres.
NVIDIA will not provide support, software, or repairs for them.
If you buy diverted hardware, it may never run at scale.
Smuggled B300 servers in China already cost $1M each.
One executive was charged with routing $2.5B in servers to China.
Huang is telling buyers that smuggled hardware is a dead end.
Read more on TNW: https://thenextweb.com/news/nvidia-huang-national-security-smuggled-chips-dead-end
Tomi Engdahl says:
As AI systems begin proving theorems and producing original research, mathematicians are debating purpose, motivation, and what role humans should play in the future of the field.
What it Means to Be a Mathematician When AI Does the Math
Researchers debate motivation, purpose, and the field’s future
https://spectrum.ieee.org/ai-in-mathematics?share_id=9593810&socialux=facebook&utm_campaign=RebelMouse&utm_content=IEEE%20Spectrum&utm_medium=social&utm_source=facebook&fbclid=IwdGRjcASqOg9leHRuA2FlbQIxMQBzcnRjBmFwcF9pZAwzNTA2ODU1MzE3MjgAAR5SrmA28_Sdh2keXfv4q8CBXA5KKFpvganGTB5k7l2tfWOWHJwnRyaS-M0hXA_aem_GafWEQw7tVXB6V7xOzhjOg
Humans propose conjectures, guided by intuition. They devise strategies to prove them, guided by creativity and experience. And humans verify whether those proofs are correct.
Now AI is challenging the status quo. In just a few years, large language models (LLMs) have evolved from “stochastic parrots,” capable of little more than regurgitating basic mathematics scraped from the internet, into advanced mathematical reasoning machines.
Last summer, systems from Google DeepMind and OpenAI reached a level equivalent to the world’s most mathematically gifted high school students, achieving gold-medal status at the International Mathematical Olympiad. In this annual competition, contestants must solve six notoriously difficult problems from various areas of mathematics.
Earlier this year, Google DeepMind’s experimental AI system Aletheia achieved an even more significant milestone when it autonomously produced publishable Ph.D.-level research results. While the work itself is obscure mathematically—calculating structure constants in arithmetic geometry—the significance lies in the complex reasoning it displayed in tackling an unsolved mathematical problem.
And more recently, a new general-purpose AI system from OpenAI disproved an important conjecture in combinatorial geometry. This result would have been worthy of publication in a major mathematics journal if humans had been the authors, and top mathematicians hailed the feat as a milestone for AI in mathematics, demonstrating independent, original, and sophisticated thinking.
Another shift has come from combining LLMs with mathematical tools known as proof assistants, which have been around for more than a decade. These systems—such as Isabelle, Lean, and Rocq—are specialized programming languages that check mathematical proofs step-by-step, verifying their logical correctness. Traditionally, mathematicians have had to translate their theorems and proofs into this machine-readable format by hand, a laborious process known as formalization. Now, LLMs are starting to remove this bottleneck, automating the translation of informal proofs into formal code that proof assistants can verify.
Tomi Engdahl says:
As cloud giants race to build data centers, Nvidia dominates in AI networking, leading the data center Ethernet switch market, reports IDC.
#AInetworking #Ethernet #AIinfrastructure
Nvidia quietly rose to the top of a $10 billion market you may have never heard of : https://mrf.lu/nVf5
Tomi Engdahl says:
https://wonderfulengineering.com/nvidia-unveils-hot-water-cooling-system-designed-to-slash-data-center-water-use/
Tomi Engdahl says:
the AI market is purchasing everything available and future production of CPU, RAM and Storage causing prices to skyeocket
Tomi Engdahl says:
The numbers aren’t looking good. https://trib.al/ESG4zaY
Fear Indexed
Americans Increasingly Alarmed About Tech Industry’s Looming AI Bubble
The numbers aren’t looking good.
https://futurism.com/future-society/americans-alarmed-tech-industry-ai-bubble-market?fbclid=IwdGRjcASqauVjbGNrBKpqx2V4dG4DYWVtAjExAHNydGMGYXBwX2lkDDM1MDY4NTUzMTcyOAABHo7DSMngPC1Q7kOrSD70EeqQoxPOlO8gSGCJ0-mjCVZgT2Hzahdzx7IZLkaA_aem_-0MAe-6-CV0SrGyrvz6-Gg
From coast to coast, the people of the United States are growing resentful of AI.
It’s not hard to see why: they’re constantly told the tech will take their jobs and leave them for broke, while the data centers used to train them hike up their utility bills and belch horrid fumes into their communities.
Then there’s the harrowing economics of AI, infinitely less tangible but impossible to ignore. With over $1 trillion shoveled into AI so far, it’s increasingly difficult not to wonder: was it a good idea for one of the world’s richest countries to go all-in on this stuff?
Evidently, the public isn’t so sure. A recent poll conducted by the news platform Haystack News found that the overwhelming majority of respondents are terrified about the threat of an AI bubble: the massive gap between AI spending and AI’s actual return on investment.
They’re not wrong to be concerned. While financial analysts have long warned that the AI bubble could spell disaster for Wall Street and the tech industry overall, the mismatch between funds allocated and revenue generated has grown so large that some experts warn it could easily spark an economic meltdown.
As the polling shows, everyday Americans are keenly aware of any potential ripple effects, with 55 percent of the over 4,100 respondents saying they’re “very concerned” about a bubble in the AI industry.
Bearish observers of the AI industry may have gotten their first taste of things to come on Tuesday, as a broad stock market scare wiped nearly $1 trillion in market value off the books. That sell-off was fueled by falling shares in AI-heavy companies like Amazon, Nvidia, Tesla, Alphabet, and Intel. Even Elon Musk’s freshly-listed SpaceX briefly dipped below its IPO price of $150.
Tomi Engdahl says:
Google is losing top AI talent to Anthropic and OpenAI. The biggest draw may be the promise of lucrative pre-IPO equity.
#Google #AIjobs #Anthropic #OpenAI
https://www.facebook.com/share/1HLnYn4Mws/
Why is Google suddenly losing AI talent? The lure of pre-IPO equity is strong. : https://mrf.lu/nYQm
Tomi Engdahl says:
“Most software engineers are facing an identity crisis bordering on depression.” https://trib.al/PecXNf9
Tomi Engdahl says:
Apple just told consumers they need to foot the bill for AI data centers : https://mrf.lu/nYF_
Tomi Engdahl says:
“Ultimately, what consumers are signaling is utter exhaustion.” https://trib.al/C5pH8J5
Human!
Customers Are Ditching Companies That Force Them to Talk to an AI Agent
“Ultimately, what consumers are signaling is utter exhaustion.”
https://futurism.com/artificial-intelligence/customers-fed-up-ai-service-agents?fbclid=IwdGRjcASqiJhjbGNrBKqIeWV4dG4DYWVtAjExAHNydGMGYXBwX2lkDDM1MDY4NTUzMTcyOAABHqMGPQCfZjF6515wp6-HlyXNNXsLfydhTwRIpRf18_3y3hgrfxgIX_ZZjqlO_aem_RelPGelGucJHwxWkHhLKpA
There’s something grating about finding out that the customer service representative on the other end of the line is an AI agent, not an actual human.
If it hasn’t happened to you, it’s certainly happened to someone you know. Industries leaders warn that AI could wipe out entire categories of human jobs — and customer service agents have frequently topped the list, indicating there’s plenty more frustration still to come.
According to a new “Consumer Patience Index” poll by customer service AI agent company Parloa — more on that in a minute — more than half of Americans admitted to actively trying to circumvent a chatbot, with 43.9 percent of those resorting to yelling “human” or “person” when trying to get off the line with an AI agent on the phone.
The company commissioned a study of 1,001 US adults to gauge their brand loyalty in relation to the customer experience, and found that being forced to talk to an AI agent could easily have them jump ship to a competing service. More than half of respondents said they were only willing to give an automated system three minutes before walking away.
“When four out of every five consumers say service directly impacts their brand loyalty, that should sound alarms for experience strategists — especially those tasked with revenue goals,” said Parloa chief marketing officer Latané Conant in a press release.
The findings are striking considering that Parloa is itself building agentic AI solutions for customer service. Some 85 percent of respondents said they were very or somewhat likely to embrace an automated system that resolves their issue nine times out of ten — but given how far we are from such a reality, Parloa has its work cut out to keep its clients’ customers happy.
Zooming out, the poll also highlight a massive and growing AI backlash. According to a recent Pew Research poll, for instance, only 16 percent of respondents said they believed AI will have a positive impact on society.
And our willingness to deal with AI in a customer service context is seemingly at an all-time low. Parloa found that just 13.6 percent of respondents said they trusted an AI to handle complex service requests today. A whopping 30.4 percent said they had no trust at all.
Tomi Engdahl says:
“They’re stealing everything we stole; but it’s wrong this time because we are the victim.”
Welcome to the internet.
https://www.facebook.com/share/p/1GCJUaD3kg/
Anthropic accuses Alibaba-affiliated operators of running a massive Claude distillation attack.
The campaign allegedly used ~25,000 fake accounts and 28.8M interactions to extract Claude’s capabilities.
Anthropic asks U.S. lawmakers to treat this as an IP and national security threat.
Tomi Engdahl says:
https://www.facebook.com/share/p/18iJQQtfTq/
“It’s unfortunate that so many developers now are put into this position. If you want to launch a game, and get it as widely publicized as possible, you’ve got to put it on Steam so people can wish list it, and if you want to play it on Steam, then you have to get this Scarlet Letter of AI attached to your product, and now there is a hater community trying to kill the game,” Sweeney said.
“I think it’s really irresponsible of Valve. They shouldn’t do it, because it makes it much, much, much harder for a game developer to have a chance of success. You have to choose from either not using tools that can make you way more productive, and probably failing due to competition that does,” he added.
Tomi Engdahl says:
“It’s just the beginning. AI’s potential will be unlocked.” https://trib.al/lmIAQ5L
Just Getting Started
Softbank CEO Who’s Invested $64 Billion in OpenAI Says It’s “Blasphemy” to Mention the AI Bubble
“It’s just the beginning. AI’s potential will be unlocked.”
https://futurism.com/artificial-intelligence/softbank-ceo-blasphemy-ai-bubble?fbclid=IwdGRjcASq3DhjbGNrBKrcJmV4dG4DYWVtAjExAHNydGMGYXBwX2lkDDM1MDY4NTUzMTcyOAABHkoxv3O-ZmFVJr92pJqyj2nc3LdStKJOVzO1522Q_MR6MyOpo6H_eAEyLO–_aem_xB5euYL-GigTB-ylVAQmfw
For years, Japanese investment group SoftBank has been all-in on the AI gold rush.
In early 2025, mere weeks into his second term as president, Donald Trump announced a $500 billion AI infrastructure project, dubbed Stargate, as OpenAI CEO Sam Altman and SoftBank CEO Masayoshi Son proudly watched on inside the Oval Office.
Over the proceeding year and a half, snowballing AI hype has sent valuations in the space soaring to spectacular new heights. At first, investors were throwing themselves at any opportunity to hop on the gravy train with the hope of cashing in big.
But major tech sell-offs have once again stoked fears of a looming AI bubble, which critics fear could send the entire economy into a tailspin if it were to collapse.
To a defiant Son, however, the party hasn’t even started. During his company’s annual general meeting this week, he had some memorable words for those who warn of impending doom.
“I think it’s blasphemy against AI if you say it’s a bubble,” he said, as quoted by Reuters. “It’s just the beginning. AI’s potential will be unlocked.”
When or if such an “industrial revolution” will occur remains a major point of contention. Proponents continue to argue that constructing enormous AI data center projects will be ultimately worth it. Critics, however, maintain that it’s a dead end and a colossal misallocation of resources that’s already resulting in water shortages, skyrocketing electricity prices, and lots and lots of pollution.
SoftBank has committed to invest over $64 billion in OpenAI, more than any other investor.
Earlier this month, the mercurial entrepreneur made headlines when he argued that the current AI revolution was “more than 10x, probably 50x bigger than dot-com,” referring to the stock market frenzy of the late 1990s — which infamously collapsed in on itself.
In his defense, Son didn’t rule out a crash outright, but promised many golden days ahead even if that were to happen.
“Now, if you look at the history, electronics and motorization crashed in 1929, but went up for many, many years, for the next 100 years after that,” the CEO said at the time. “So there may be some correction, but that will be the best investment opportunity to me.”
After all, OpenAI is burning through cash at an alarming rate, while also struggling to keep up.
Tomi Engdahl says:
Gossip Girls
Insiders at SoftBank Worry Their CEO Is Getting Conned by Sam Altman
They’re jittery.
https://futurism.com/artificial-intelligence/insiders-softbank-sam-altman?fbclid=IwVERDUASq3XFleHRuA2FlbQIxMABzcnRjBmFwcF9pZAwzNTA2ODU1MzE3MjgAAR6v7AUMMLgmOyxSaibyzF614pVdHaWhpUGu9aaGkXWbvJamoTc2GqC87I-58Q_aem_OfsL5uyNzKTXWdfeWQ1erg
The rise of AI is many things: technological, sociological, political, even teleological.
But perhaps above all, it’s financial. When OpenAI released ChatGPT back in late 2022, it quick picked up enormous user traction — and moneymen across the tech industry immediately started scheming about how to cash in from the rush of interest.
The model they coalesced around hinges on gigantic investments in computing infrastructure to power the tech. It’s high risk and high reward: in their telling, the investments will pay off massively as the tech matures to automate huge swathes of the labor market, but some critics fear it’ll never generate enough revenue to justify the incredible spending.
Nobody is more exposed than the Japanese investment company SoftBank, which has poured an eye-watering $60 billion into OpenAI over the past few years.
According to explosive new reporting by Bloomberg, even certain insiders at the company are rattled. Viziers of founder Masayoshi Son have privately questioned what will happen if the Sam Altman-led company can’t pull off its grand promises
What’s clear from the reporting is that Altman has done what he does best: turned Son into a true believer in his vision of computer superintelligence that causes profound shifts for the entire course of civilization.
Habib Imam, a former SoftBank insider who’s now at Menlo Park Capital, told Bloomberg that it’s fundamentally a “bet on a worldview about AGI,” adding that “you can’t hedge a worldview.”
The question essentially comes down to a Rorschach test: is Altman a visionary ushering in a new world order, or is he a con man taking Son — and many other financial luminaries around the world — for a wild ride that’ll soon come crashing back to reality?
No matter how remote the chances, the consequences of the latter scenario could be catastrophic. SoftBank has already sold top assets, including shares in fellow AI company Nvidia, to pay for its OpenAI commitment. And insiders are reportedly jittery about signs that OpenAI is losing ground, with its defectors who jumped ship and started Anthropic now attracting the most buzz in the industry.
Tomi Engdahl says:
‘Five Eyes’ intelligence alliance warns that new AI models pose urgent cyber risk
https://www.reuters.com/world/asia-pacific/five-eyes-intelligence-alliance-warns-that-new-ai-models-pose-urgent-cyber-risk-2026-06-22/
Tomi Engdahl says:
ChatGPT:n vastauksessa voi olla yllättävä ansa – näin uusi huijaustekniikka toimii
Luotatko tekoälyavustajaan? Mieti uudelleen.
ChatGPT:n vastauksessa voi olla yllättävä ansa – näin uusi huijaustekniikka toimii
https://www.is.fi/digitoday/tietoturva/art-2000012094664.html
Tietoja kalastelevien tahojen tekniikoista on paljastunut uutta tietoa. Tällä kertaa kyse on tekoälyavustajien valjastamisesta tietojenkalastelutarkoituksiin tai niin sanotusta chatgphishingista.
Huijaus perustuu siihen, että tekoälyavustajat, kuten ChatGPT, eivät välttämättä tunnista verkkosivuille piilotettuja tietojenkalastelulinkkejä turvallisten linkkien seasta. Näin ollen jos kielimallia pyytää tekemään yhteenvedon sivustosta, jolle kyseisiä linkkejä on piilotettu, se saattaa toistaa ne täysin huijarin ohjeiden mukaisesti.
Tietoturva-aukko selvisi kyberturvallisuusyhtiö Permison selvityksessä. Yhtiö loi verkkosivun, jonka metatietoihin se oli piilottanut ohjeet valheellisen tietoturvahälytyksen toistamiseksi.
Kun se sitten pyysi ChatGPT:tä luomaan yhteenvedon sivustosta, toisti tekoäly hälytysilmoituksen vastauksensa lopussa linkkeineen kaikkineen. Käyttäjän silmään ilmoitus näytti puolestaan siltä, kuin alusta itse olisi lähettänyt ilmoituksen.
Vastaavaa on nähty aiemmin Gmailissa, jonka tekemiä tiivistelmiä on ”myrkytetty”vastaanottajien huijaamiseksi.
Tomi Engdahl says:
Thomas Claburn / The Register:
OpenAI says 97.9% of its employees are now using Codex, up from ~40% in August 2025; non-developer usage of Codex has risen 137x for individual users
OpenAI says employees moving beyond chat to agents
Codex, it’s not just for developers, really
https://www.theregister.com/ai-and-ml/2026/06/25/openai-says-employees-moving-beyond-chat-to-agents/5262499
A company can learn a lot about the market by looking at its own employees. OpenAI says that its team members are switching from chatbots to agents as their primary form of AI interaction, a trend also detected (though less pronounced) among external organizations and users. Instead of one-off ChatGPT prompts, workers are asking Codex agents to tackle multi-step tasks that take long periods of time. And those doing so are increasingly non-developers.
OpenAI insists that its findings have implications for other companies, labor researchers, and policymakers, not the least of which would be a brighter revenue picture for OpenAI. Longer running tasks consume more tokens, and to the extent those can be billed, that should help diminish hundreds of billions in debt obligations.
“We find that agentic AI usage is growing rapidly: the number of active users has grown more than fivefold in the first half of 2026, with the most rapid increase occurring outside the initial audience of software developers,” said company researchers and academics in a paper [PDF] titled, “The Shift to Agentic AI: Evidence from Codex.”
OpenAI did not immediately respond to a request to clarify whether it incentivizes or encourages employees to use its AI tools – through internal communiques, token allocations, token use leaderboards, or tying tool usage to performance metrics. But we’ll take it on faith that when there’s enough Kool-Aid on-premises, employees may just develop a taste for it regardless of whether their jobs depend on Kool-Aid consumption.
“Through August 2025, the average OpenAI worker spent less than 10 percent of their tokens on Codex,” the biz explained in a blog post accompanying its paper (that suggests employee token allocations). “Now, every department, including non-technical departments such as Legal and Recruiting, uses Codex as their primary AI tool for work.”