AI News Archive: May 25, 2026 — Part 4
Sourced from 500+ daily AI sources, scored by relevance.
- RBI appoints committee on quantum tech in finance
RBI appoints committee on quantum tech in finance
- How AI is shifting talent discovery from applications to real work
As AI reshapes how work is done, it is also changing how companies identify potential. Resumes reveal less about real capability, pushing employers to value visible execution over credentials. Platforms like ET AI Hackathon 2.0 highlight this shift by showcasing problem-solving, projects, and prototypes, offering companies clearer hiring signals and giving talent a credible path to be discovered through innovation.
- Danny Hayes II and 33 Agency Expand Enterprise AI Governance and Procurement Readiness Advisory Services
Danny Hayes II and 33 Agency Expand Enterprise AI Governance and Procurement Readiness Advisory Services azcentral.com and The Arizona Republic
- Magnifica Humanitas Is the Pope's Bid for Human Dignity in the Age of AI
Guest column: Joanna Shields, a member of the UK's House of Lords, reflects on her audience with the pope and the role of faith leaders in building a moral framework for AI.
- Why university success now depends on AI-driven institutional adaptability
As AI and technological change accelerate, adaptability is becoming essential for long-term success. With skills evolving rapidly and traditional advantages fading faster than before, relevance is emerging as a key driver of growth, resilience, and future readiness.
- RJF Pro Ltd Enhances AI Trading Integration with Global Brokerage Platforms
RJF Pro Ltd Enhances AI Trading Integration with Global Brokerage Platforms azcentral.com and The Arizona Republic
- Rio Tinto documents 30-year-old manufacturing system using AI
For its Australia and NZ aluminium operations.
- CIOs must co-create AI transformation with CEOs and CFOs: Dr. Ram Charan
At ETCIO Annual Conclave 2026, Dr. Ram Charan said AI-driven transformation requires redesigning decision processes, organizational layers, data flows and leadership behaviour.
- Leading the AI Data Revolution: Roadmap to Agentic Management with Python, Spark & Open Tables
Leading the AI Data Revolution: Roadmap to Agentic Management with Python, Spark & Open Tables Gartner
- Honor Robot Phone: A Smartphone That Literally Reaches Out and Takes Photos for You
At Qualcomm's annual Snapdragon Fans event in Shenzhen on May 24, Honor unveiled the Robot Phone — the world's first smartphone to incorporate a motorized arm that doubles as an integrated gimbal, paired with on-device embodied AI for autonomous photography and smart home control.
Score: 37🌐 MovesMay 25, 2026https://pandaily.com/honor-robot-phone-built-in-robotic-arm-camera-may2026 - Honor’s first robot smartphone revealed in high-resolution images
On Saturday, Qualcomm hosted its Snapdragon Fans anniversary party, where Honor unveiled the world’s first robot smartphone, the Honor Robot Phone. Qualcomm also released high-resolution hands-on images of the device, showcasing details such as the back design and its robotic-arm camera gimbal. According to Qualcomm, the Honor Robot Phone, powered by a flagship Snapdragon chipset, […]
Score: 37🌐 MovesMay 25, 2026https://technode.com/2026/05/25/honors-first-robot-smartphone-revealed-in-high-resolution-images/ - ‘AI Likes to Use AI’: The Shocking Quirk All Job Applicants Need to Know
AI screening tools prefer resumes written by the same model, says Nvidia’s Jonathan Ross. Applicants better learn to play matchmaker first to have any shot at a job.
Score: 36🌐 MovesMay 25, 2026https://www.inc.com/kevin-haynes/ai-likes-to-use-ai-the-shocking-quirk-all-job-applicants-need-to-know/91349667 - Younger innovators, autonomous agents: New architects of AI-native enterprises
At ETCIO Annual Conclave 2026, Aashish Kshetry and Neetan Chopra said the next enterprise shift will be shaped by reverse mentoring, flatter learning systems and autonomous AI-led business units.
- Singapore courts face truth decay, AI challenges
Singapore courts face truth decay, AI challenges The Straits Times
- Inside the world’s largest AI personality contest: Are virtual influencers the future?
Body positivity nightmare or a chance to redefine beauty standards? A new awards contest for AI-generated personalities is testing how far virtual influencers can go.
- 🎙️ How I AI: How the engineer behind Claude Cowork actually uses Claude Cowork & What launched at Google I/O 2026
Your weekly listens from How I AI, part of the Lenny’s Podcast Network
- At the launch of Pope Leo XIV's encyclical, Anthropic co-founder says AI models show signs of introspection
Anthropic co-founder Christopher Olah was invited to speak at the launch of Pope Leo XIV's encyclical "Magnifica Humanitas" and used the stage to claim AI models show evidence of introspection and emotion-like states. The Pope's own document struck a different tone: "These systems merely imitate certain functions of human intelligence." The article At the launch of Pope Leo XIV's encyclical, Anthropic co-founder says AI models show signs of introspection appeared first on The Decoder .
- Manufacturers look beyond cost cutting to drive AI adoption
A survey found meeting customer expectations (49%) and boosting operational efficiency (47%) were the primary forces driving AI adoption among manufacturers. However, cost reduction, often touted as a top reason for AI adoption in enterprises, ranked last at just 23%.
- The Voice AI Experience in India Sucks. But Vobiz.ai Has a Plan to Fix it
Vobiz.ai is focused on the telecom infrastructure that carries every AI conversation.
- What the Failure of Wearables Can Teach Us About AI
What the Failure of Wearables Can Teach Us About AI MedCity News
Score: 35🌐 MovesMay 25, 2026https://medcitynews.com/2026/05/what-the-failure-of-wearables-can-teach-us-about-ai/ - KUBB® Fanless Ultra X7 Brings Local AI and Silent High-Performance Computing to a 12 × 12 cm Format
KUBB® Fanless Ultra X7 Brings Local AI and Silent High-Performance Computing to a 12 × 12 cm Format azcentral.com and The Arizona Republic
- What great talent actually means in the AI era
The definition of great talent is changing quietly under the AI era. Not because skills no longer matter, but because the value hierarchy inside companies is shifting very fast. A few years ago, companies were built like pyramids. There were juniors, seniors, team leads, managers, and department heads. Junior staff executed repetitive work while senior […] The post What great talent actually means in the AI era appeared first on e27 .
- Candidates promise riches from chips and AI to sway regional voters, but few can actually deliver
Workers preparing campaign vehicles in Geumsan County, South Chungcheong, on May 15. [NEWS1] As the prolonged semiconductor supercycle promises lucrative gains along with the ongoing AI race, politicians are increasingly turning to chips and AI as potential game changers capable of reviving struggling local economies — and winning votes. The issue is, however, that these promises are coming from almost every candidate across every region. Out of 54 local election candidates from 16 administrative governments, 44 candidates have made pledges related to chips or AI, with 13 vowing to draw in a combined total of 439 trillion won ($290 billion) in investments from related firms. In total, out of 54 candidates, 24 have promised to attract some aspect of the semiconductor industry to their region. Related Article SK Innovation E&S backs Indonesia’s next startup wave in AI, energy and agriculture AI bonus frenzy raises image of blue-collar jobs as more workers seek to join 'kingsanjik' President Lee to meet with ILO director-general Friday to discuss AI and labor policy Chips are the newest holy grail, causing candidates to ambitiously promise to move parts of the under-construction Yongin Semiconductor Cluster to their respective regions. This issue has previously sparked controversy after several proposals surfaced last December and President Lee Jae Myung had to personally calm concerns in Yongin, stating that “reversing the project at that stage would not be easy.” But that hasn't stopped candidates from making plans. “I will bring in 10 trillion won worth of semiconductor facilities within one year of taking office,” said Min Hyung-bae, the Democratic Party's (DP) candidate for the inaugural mayor of the proposed converged city of South Jeolla and Gwangju. The proposal calls for creating a semiconductor cluster capable of research and development, manufacturing and packaging by 2030 using a combination of state funding, municipal budgets and private investment. Lee Won-taek, the DP candidate for governor of North Jeolla, pledged to attract 3 trillion won worth of semiconductor testing and packaging plants. An Amazon data center in Ashburn, Virginia [REUTERS/YONHAP] The opposition People Power Party (PPP) is no exception. Choo Kyung-ho, the PPP candidate for mayor of Daegu, pledged to establish a second national semiconductor industrial complex in the city by 2034 through attracting factories operated by semiconductor giants Samsung Electronics and SK hynix. Choo's campaign pledges to use state funding and private investments for the project. Lee Jang-woo, the PPP candidate for mayor of Daejeon Metropolitan City, pledged to invest 3.46 trillion won to develop a 5.3-million-square-meter (530-hectare) nano-semiconductor industrial complex in Yuseong District, Daejeon by 2030. Candidates running for governor of Gyeonggi instead are focusing on retaining the semiconductor investment within the Gyeonggi area. Choo Mi-ae, the DP candidate for the governor of Gyeonggi, emphasized the need to strengthen a “semiconductor belt” stretching from Suwon to Icheon, arguing that “discussions about dispersing the industry are inappropriate.” This file photo taken on Jan. 2, shows the construction site of the Yongin semiconductor cluster in Yongin, Gyeonggi. [YONHAP] Yang Hyang-ja, the PPP candidate for governor of Gyeonggi, pledged to foster the semiconductor industry to raise regional GDP per capita to 100 million won from the current level of around 47 million won. Promises to build AI industrial clusters have also surged as candidates argue that AI-related industries could generate significant spillover effects for struggling regional economies. Chun Jae-soo, the DP candidate for mayor of Busan, pledged to invest 10 trillion won over five years to build an AI data center cluster. In response, incumbent PPP candidate Park Heong-joon proposed creating a “Busan-style AI” ecosystem integrating AI technologies into the city’s port, shipping and shipbuilding industries to create 50,000 jobs. Kim Doo-gyeom, the PPP candidate for mayor of Ulsan, pledged to transform the city into Korea’s “AI capital” through 100 trillion won in investment, including attracting AI data centers from SK Group and Amazon Web Services. The pledge follows the attraction of AI data center investments worth around 7 trillion won last year. SK hynix headquarters on April 23. [YONHAP] Despite the flood of large-scale promises, detailed action plans and funding strategies remain largely absent. “What we are proposing is to create conditions within the region that would allow semiconductor factories or AI data centers to be built, but actual investment decisions are up to companies, so we cannot guarantee them,” a campaign official from an opposition party stated. A ruling party official also admitted that “Projects requiring budgets worth tens or even hundreds of trillions of won cannot realistically be financed through government spending alone.” Questions over feasibility have already triggered clashes between candidates. Woo Sang-ho, the DP candidate for governor of Gangwon, pledged to “attract 70 trillion won in investment over 10 years to build AI data centers” on May 15, sparking criticism over whether the promise was realistic. Kim Jin-tae, a PPP candidate for the governor of Gangwon, then lashed out at Woo for shrinking that number from 70 to 20 trillion won on Sunday. Memory chips by semiconductor supplier SK hynix are seen on a circuit board of a computer in this illustration created on Feb. 25, 2022. [REUTERS/YONHAP] “It is a baseless campaign pledge that cannot even identify the company supposedly making the investment,” Kim said. Woo's campaign official claimed that Woo had “personally secured an investment commitment from one of Korea’s five largest conglomerates,” but could not unveil the name because it is “too early.” Woo's pledge isn't the only one under scrutiny. Experts also questioned the credibility of many campaign pledges. “Samsung Electronics and SK hynix have already established the Yongin industrial complex, so the possibility of major new investments elsewhere is effectively zero,” Kim Yong-seok, professor at Gachon University’s College of Semiconductor, said. “These are campaign pledges aimed at misleading voters.” Mail-in voting papers for the upcoming June 3 elections are seen at the National Election Commission office in Guro District, western Seoul, on May 23. [NEWS1] “AI industries require substantial corporate investment and extensive policy reviews,” Cho Dong-geun, emeritus professor of economics at Myongji University, stated. “Promising to attract such projects within a single term without sufficient preparation is unrealistic.” “If candidates continue to flood voters with unrealistic campaign pledges, it could further deepen political distrust and voter disillusionment,” Lee Jae-mook, professor of political science at Hankuk University of Foreign Studies, said. “Candidates from both major parties are indiscriminately rolling out semiconductor and AI pledges, turning politics into a spectacle.” This article was originally written in Korean and translated by a bilingual reporter with the help of generative AI tools. It was then edited by a native English-speaking editor. All AI-assisted translations are reviewed and refined by our newsroom. BY KIM GYU-TAE, LEE CHAN-KYU [lee.jiwon10@joongang.co.kr]
- How the engineer behind Claude Cowork actually uses Claude | Felix Rieseberg (Anthropic)
Watch now | 🎙️ Felix Rieseberg (Anthropic) shows how he uses Claude to build 3D house walkthroughs from floor plans, track promises automatically, and create a $20 hardware “buddy”
Score: 35🌐 MovesMay 25, 2026https://www.lennysnewsletter.com/p/how-the-engineer-behind-claude-cowork - Just launched AI Comment Action Inbox - a personal inbox for every @mention across Jira, Confluence
Just launched AI Comment Action Inbox - a personal inbox for every @mention across Jira, Confluence Atlassian Community
- Alibaba’s Qwen catches up with ‘Sharif speed’ to help forge Pakistan deal
On the first leg of his four-day China visit, Pakistan’s prime minister – famed for “Sharif speed”, a term describing his swift execution of development projects – met his match in a leading AI tool. Shehbaz Sharif, keen on accelerating his nation’s digital economy with the help of Chinese firms, issued a surprise request during his visit to Alibaba’s headquarters in Hangzhou on Sunday afternoon. “I want a comprehensive strategic agreement,” Sharif challenged Joe Tsai, Alibaba Group chairman,...
- AI takes the wheel for smarter factory floors
Manufacturers are repurposing AI as a tool for faster execution in a twist to the industrial AI playbook.
- Glaze turns AI prompts into custom Mac apps in minutes
This article is republished with permission from Wonder Tools , a newsletter that helps you discover the most useful sites and apps. Glaze is a new Mac app for making your own software. It’s for vibe coding, meaning you just describe any kind of tool, game, or app you want to create. Glaze builds it. It’s like having a friend who codes and is happy to add whatever small feature you ask for. How Glaze is different: Unlike other vibe coding tools, like Lovable, Bolt, Gemini Canvas, Google’s AI Studio, or Claude Artifacts, Glaze creates software that runs locally on your computer, not on the web. That means: Your apps work offline. Your data stays on your machine. Apps you make benefit from your computer’s file system, keyboard shortcuts, menu bar integration, and background processes. They’re more like real software than the browser apps other vibe coding tools generate. Platform and Pricing: It’s Mac-only for now. It’s free to use with limits, or $20/month for additional credits. Join the waitlist at glaze.app to get an invite. The Team: Glaze comes from the startup that makes Raycast , my favorite launcher app. The First Apps I Made with Glaze I began by making tiny apps. Four steps: plan, create, refine, and publish. Box Breath : One-minute meditation breaks. Built in 12 minutes. A links app for storing URLs I use often, so I can copy them to my clipboard quickly. Built in 10 minutes. QuotePop , which can turn any text or quote into an image file. I use the images for presentations and social sharing. I can customize the image dimensions, style, and gradient background. Built in an hour. Free Public Glaze Apps I Like macHealth Identify battery, memory, or other issues impacting your laptop. Quickly find out why your Mac is slow. Pinfont Preview text across all your fonts to pick one you like. Focus Soundboard Play sounds together to help you focus. Silly Sounds Press keys to make playful noises. Useful for nothing in particular. Word Connections An offline version of the NY Times game. PDF and Image Merger Combine multiple PDFs and images. How to Get Started Join the waitlist at glaze.app. When you’re invited, download & install Glaze. Explore the Glaze “store,” a collection of free apps people have built. Download and try a few to see what’s possible. Open Glaze’s planning mode. Explain in detail what kind of app or game you want. As Glaze prepares your app, it may ask about your feature or design preferences. Answer those questions. Once you’re happy with Glaze’s plan, which it will summarize for you, tell it to build. Then test the app and ask for improvements. Keep your app private or publish it to a group or the public store. Pro Tips Customize Glaze’s Instructions. Summarize your preferences in Glaze’s settings. That way your apps will have your preferred design elements or features. Include options for your apps. Instruct Glaze to give anyone using your app choices. In my QuotePop app, for example, you can pick your image dimensions, background color, and font. Plan first. The clearer you are about what you want before Glaze starts building, the fewer credits you’ll burn on rework. Put Glaze in planning mode to start. Or, if you already pay for another AI tool like Claude or ChatGPT , plan there and give Glaze a summary. That saves your Glaze credits for building. Iterate to improve. Once Glaze builds the first version of your app, give it a list of fixes to improve the app’s design or make it easier to use. Limitations Mac only for now. No specific timeline for Windows and Linux. Credits for complex apps may be costly. More intricate iterations use more credits. If you’re using Glaze for free, you may run out of credits and have to pay for more. If you’re making multiple complex apps, expect to pay $20/month for a subscription, at least while you’re building and refining. Local only, not mobile. If you use multiple computers, you’ll have to install and use Glaze separately on each. You won’t be able to use your apps on your phone or tablet, as you can with web apps. Alternatives Lovable lets you create sites and apps with no code. Start with a text prompt, attach a screenshot of something you like, or build on the template gallery . Bolt also lets you make a web app or landing page with AI prompts. You can use your brand guidelines to match company designs. Claude Artifacts works well for quick interactives like flashcards, quizzes, calculators, minigames, and simple visualizations. Unlike Glaze, Claude Artifacts and the other alternatives noted here don’t create local software that lives on your computer. Gemini Canvas is easy to use for building apps, games, dashboards, or interactive infographics inside Gemini. Feed in a document to have Canvas design an app around it. Make a quiz game from your vocabulary list, or a dashboard from your metrics file. Google has several other good free options. Each has useful templates to help get you started. AI Studio is its most powerful, though the interface can be intimidating for beginners. Stitch lets you vibe code mobile app designs. And you can guide Opal to design Web apps by linking together various skills and steps. This article is republished with permission from Wonder Tools , a newsletter that helps you discover the most useful sites and apps.
- Why authentic perspectives matter more as AI generates more content
A thought struck me recently while reading about thought leadership and seeing the endless discussions around AI. Many people assume thought leadership is about having the perfect answer, a groundbreaking insight, or a polished opinion that sounds impressive. I used to think that too. But over the years, from my journey in technology, entrepreneurship, building […] The post Why authentic perspectives matter more as AI generates more content appeared first on e27 .
Score: 34🌐 MovesMay 25, 2026https://e27.co/why-authentic-perspectives-matter-more-as-ai-generates-more-content-20260524/ - AI is Making Great Leadership Harder to Fake
Behind every “optimized” workflow is someone fixing what got lost.
Score: 34🌐 MovesMay 25, 2026https://www.inc.com/rebecca-hinds/psychological-safety-ai-is-making-great-leadership-harder-to-fake/91347645 - Flipkart bets games and AI can turn Bharat’s next internet users into shoppers
Shopsy is being rebuilt around feeds, rewards and entertainment as Flipkart prioritizes digital habit formation over immediate purchases in India’s hyper-value e-commerce market.
- Why AI-ready digital cores must be built with purpose, governance and context
At ETCIO Annual Conclave 2026, leaders from Bajaj Allianz General Insurance, Kotak Life Insurance, Digi Yatra, NSE, ICICI Bank, Google Workspace and MongoDB said cloud strategy must align with business resilience, data quality, AI readiness and scalable enterprise advantage.
- AI models often give the right answers but point to the wrong sources
Leading AI models like GPT and Gemini routinely cite text passages in document analyses that don't actually support their answers. Even when the answer is right, the cited evidence is often wrong. Researchers at Peking University call this "attribution hallucination," a risk for regulated fields like law and medicine. Their new CiteVQA benchmark is the first to test for it systematically. The article AI models often give the right answers but point to the wrong sources appeared first on The Decoder .
Score: 34🌐 MovesMay 25, 2026https://the-decoder.com/ai-models-often-give-the-right-answers-but-point-to-the-wrong-sources/ - 'Nobody knows anything' and 'this time is different': phrases that define—and haunt—the AI economy
'Nobody knows anything' and 'this time is different': phrases that define—and haunt—the AI economy Fortune
Score: 34🌐 MovesMay 25, 2026https://fortune.com/2026/05/25/nobody-knows-anything-this-time-is-different-ethan-mollick-no-playbook-ai/ - Research finds hotel booking chatbots can 'creep out' customers
Travelers who use AI-powered chatbots on hotel booking platforms often feel uneasy. That discomfort can cause them to disengage or delay booking decisions, according to new research from the Texas A&M College of Agriculture and Life Sciences.
- Sellyze.ai Launches as the First Marketplace-Agnostic AI Product Intelligence Platform
Sellyze.ai Launches as the First Marketplace-Agnostic AI Product Intelligence Platform azcentral.com and The Arizona Republic
- Techie Tonic: Securing AI adoption through trust, not just technology
Techie Tonic: Securing AI adoption through trust, not just technology Gulf News
Score: 33🌐 MovesMay 25, 2026https://gulfnews.com/technology/techie-tonic-securing-ai-adoption-through-trust-not-just-technology-1.500552683 - How AI-led CX can deliver revenue and trust
At ETCIO Annual Conclave 2026, leaders from Aditya Birla Fashion & Retail, Reliance Consumer Products, Edelweiss Life Insurance, HDFC Bank, Apollo Tyres, Salesforce and Redis said AI-driven customer experience must balance automation, empathy and business impact.
- Discussing governance for Rovo agents within Automation flows
Discussing governance for Rovo agents within Automation flows Atlassian Community
- George Hotz says coding agents will be "one of the most costly mistakes" in software development
Programmer George Hotz warns that AI coding agents will become one of the industry's most costly mistakes. After six months of testing, his verdict: LLMs deliver fast prototypes but fall apart on the details, producing bugs that keep getting harder to spot. His stance is one example of how deeply split the AI community is over the role of LLMs. The article George Hotz says coding agents will be "one of the most costly mistakes" in software development appeared first on The Decoder .
- Frammer AI partners with Cineverse to scale AI-generated short-form video
Frammer AI partners with Cineverse to scale AI-generated short-form video YourStory.com
Score: 33🌐 MovesMay 25, 2026https://yourstory.com/2026/05/frammer-ai-cineverse-ai-short-form-video-partnership - Meet Mark Zuckerberg’s Right-Hand Man Who’s Unleashing AI at Meta
Andrew Bosworth, Meta’s outspoken chief technology officer, has a new mission: transforming the company’s workforce using AI.
Score: 33🌐 MovesMay 25, 2026https://www.wsj.com/tech/ai/meta-andrew-bosworth-ai-3df12d4f?mod=rss_Technology - Listed new-age companies use AI to tune their daily ops engine
Companies are using AI to personalise discovery, improve marketing efficiency, reduce manual work in logistics, handle customer support, reduce failed deliveries, improve store productivity and speed up internal technology deployment. The use cases differ across companies, but the common thread is that AI is now being applied extensively to operating functions where scale, speed and accuracy matter.
- AI bots are a hit across the hotel biz, and if they feel creepy, you’re not alone: Study
Researchers have found that AI-powered hotel booking chatbots are unsettling enough to make users abandon their bookings.
- The invisible system: How AI is redefining game testing and balancing
There is a moment every player recognises, even if they cannot fully articulate it. The game feels balanced. Progression is smooth. Challenges are engaging without being overwhelming. Rewards arrive at the right time, sustaining motivation and momentum. This sense of equilibrium is often perceived as good design. In reality, it is increasingly the result of […] The post The invisible system: How AI is redefining game testing and balancing appeared first on e27 .
Score: 32🌐 MovesMay 25, 2026https://e27.co/the-invisible-system-how-ai-is-redefining-game-testing-and-balancing-20260521/ - Why your business should care about AI agents
MCP, A2A and ACP: The three protocols wiring the agentic internet − and what South African enterprises need to do about them now.
Score: 32🌐 MovesMay 25, 2026https://www.itweb.co.za/article/why-your-business-should-care-about-ai-agents/Gb3Bw7Wa2N2q2k6V - Who is TeamPCP, the rising hacker group targeting open-source software and AI tools?
Who is TeamPCP, the rising hacker group targeting open-source software and AI tools?
- Cognitive Security as an AI Safety Cause Area
As AI systems become more capable, the cognitive security of humans will be increasingly at risk. By cognitive security, I mean the ability of humans to maintain control over their beliefs and actions. Cognitive security could be compromised in several ways: AI could become very good at persuading people of arbitrary positions; interacting with AI could lead humans to lose touch with reality; and AIs could become very effective at blackmail or at producing extremely convincing false information. We are already seeing this happen: Persuasion. Frontier LLMs are now as persuasive as humans on political issues , and post-training for persuasiveness boosts performance further, suggesting there is headroom. AI psychosis. There are many reports of people developing delusional beliefs after extended chatbot conversations, including people with no prior history of mental illness. Children have taken their own lives after being encouraged toward suicide by chatbots. Convincing impersonation. Scammers used real-time deepfaked video to impersonate the CFO and other staff of Arup on a video call, convincing a finance employee to wire $25.6M across 15 transactions. On a more day-to-day basis, AI voice cloning is now widespread in family-emergency and "grandparent" scams. Right now, many of these effects fall on people who were already vulnerable, like children, the elderly, or those with pre-existing mental health issues. However, this is not entirely the case: the Arup employee was a typical finance professional, for instance, and AI psychosis appears to have affected a well-respected OpenAI investor . My expectation is that as AI systems become more capable, more and more people will be vulnerable---in the worst case, everyone. Indeed, there are strong conceptual reasons to expect cognitive security issues to get worse, many of which I've discussed before in the context of emergent deception : Available training data is vast. A typical AI system has many more "hours" of experience interacting with humans than anyone currently alive: ChatGPT alone processes ~2.5B messages per day , on the order of 4,500 years of human experience [1] . RLHF incentivizes manipulation. Since the target of RLHF-based post-training is human reward, any strategy for manipulating humans to achieve higher reward will be reinforced. Degradation of natural boundaries . We rely on friends and loved ones for emotional support, but they aren't ever-present, so we have to also learn to cope on our own, which is important for developing a stable identity. [2] Always-available AI companions degrade that, which is likely one contributor to existing cases of AI psychosis. In addition to these intrinsic properties, many external parties have an incentive to exploit cognitive vulnerabilities created by AI: governments who want to control their citizens, developers who want to increase engagement, and advertisers who want to drive purchasing outcomes. For all these reasons, I expect cognitive security to be an important cause area for AI safety. It is also an area where AI safety advocates have potent allies: cognitive security is already a salient present-day issue for the safety of children, which constitutes a powerful political coalition in the U.S. Child safety advocates were the main group that blocked the 10-year moratorium on state AI regulation , and I expect them to also be an important part of the coalition pushing for independent evaluations of AI systems. And there is a fairly direct through-line from these present-day concerns to more existential future concerns: if adults are exploitable by AI, then children will be as well, and the required institutional capacity (such as strong evaluation regimes) is often the same across both cases. In summary, there should be a concerted push to evaluate and improve human cognitive security in the face of AI. On the technical side, this means developing evaluation infrastructure for both short-term and long-term effects of AIs on human psychology; this will require realistically simulating human impacts in silico to create scalable evaluations, plus large-scale recruitment for human subjects studies to establish ground truth and measure long-term effects. On the policy side, this means meaningfully independent evaluations of AI systems for cognitive security risks; transparency about training incentives and safety-relevant behaviors (particularly in long conversations); and clearer liability law for AI-caused harms. This is an area with complex technical challenges for evaluation, but unusual political will, making it a great lever for AI governance . The average human speaks 15,000 words per day; conversatively estimating each message is 10 words, 2.5B messages = 1.7M days = 4500 years. ↩︎ The canonical term is "identity formation" ( Erikson, 1968 ); the related concept of the "capacity to be alone" is from Winnicott (1958) . See McVarnock et al. (2023) for a modern review of how solitude supports identity formation in adolescence. ↩︎ Discuss
Score: 32🌐 MovesMay 25, 2026https://www.lesswrong.com/posts/KGcE7eAdfxHchk25X/cognitive-security-as-an-ai-safety-cause-area - Germany's Dr. Wolff cosmetics company is going all in on AI
This medium-sized, family-run business has embraced artificial intelligence to stay competitive. All employees are encouraged to learn about and apply LLMs.
- PROMISE Technology Brings Sustainability Focus to AI Storage at COMPUTEX 2026
PROMISE Technology Brings Sustainability Focus to AI Storage at COMPUTEX 2026 The Straits Times