OpenAI is awarding up to $2 million in grants for research at the intersection of AI and mental health. The program supports projects that study real-world risks, benefits, and applications to improve safety and well-being.
Today we’re announcing new funding—$40B at a $300B post-money valuation, which enables us to push the frontiers of AI research even further, scale our compute infrastructure, and deliver increasingly powerful tools for the 500 million people who use ChatGPT every week.
We’ve written a policy research paper identifying four strategies that can be used today to improve the likelihood of long-term industry cooperation on safety norms in AI: communicating risks and benefits, technical collaboration, increased transparency, and incentivizing standar…
We’ve written a paper arguing that long-term AI safety research needs social scientists to ensure AI alignment algorithms succeed when actual humans are involved. Properly aligning advanced AI systems with human values requires resolving many uncertainties related to the psycholo…
arXiv:2604.24618v1 Announce Type: new Abstract: We evaluate the propensity of frontier models to sabotage or refuse to assist with safety research when deployed as AI research agents within a frontier AI company. We apply two complementary evaluations to four Claude models (Mytho…
We evaluate the propensity of frontier models to sabotage or refuse to assist with safety research when deployed as AI research agents within a frontier AI company. We apply two complementary evaluations to four Claude models (Mythos Preview, Opus 4.7 Preview, Opus 4.6, and Sonne…
Artificial intelligence now decides who receives a loan, who is flagged for criminal investigation, and whether an autonomous vehicle brakes in time. Governments have responded: the EU AI Act, the NIST Risk Management Framework, and the Council of Europe Convention all demand tha…
METR (Model Evaluation & Threat Research)
TIER_1English(EN)·
<p style="text-align: center;"><a class="button button-primary button-wide max-width-100" href="https://metr.org/frontier-ai-regulations.pdf">View as PDF</a></p> <p>Frontier AI developers such as OpenAI, Google, Anthropic, xAI, and others are governed by safety and security oblig…
METR (Model Evaluation & Threat Research)
TIER_1English(EN)·
<p>A number of developers of large foundation models have committed to corporate protocols that lay out how they will evaluate their models for severe risks and mitigate these risks with information security measures, deployment safeguards, and accountability practices. Beginning…
METR (Model Evaluation & Threat Research)
TIER_1English(EN)·
<p><span>With the release of Claude Mythos, it feels like we are approaching the end-game of AI safety, where the number of parties that can make a real impact shrinks down to the handful of labs at the frontier, a few companies too critical to exclude from the conversation, and …
<p><span>In February, the </span><a href="https://www.swiftcentre.org/"><b><span>Swift Centre for Applied Forecasting</span></b></a><span> launched a competition designed to bridge the gap between abstract AI safety research and the realities of government decision-making. </span…
<p><span>Hello! I have found myself in a position that I think could benefit AI safety/alignment, and I figured I would ask here for suggestions on how to most effectively use it.</span><br /><br /><span>I am eployed at a top 25 public research university. I don't really have a o…
<p><span>We are organizing a symposium on the intersection of algorithmic information theory and machine learning July 27-29th at Oxford!</span></p><p><span>See the announcement here for details: </span><a href="https://sites.google.com/site/boumedienehamzi/third-symposium-on-mac…
Future of Life Institute
TIER_1English(EN)·Chase Hardin·
The Future of Life Institute has released its first safety scorecard of leading AI companies, finding many are not addressing safety concerns while some have taken small initial steps in the right direction.
The Decoder
TIER_1English(EN)·Maximilian Schreiner·
<p><img alt="Amazon is introducing automatic prompt optimization for its Bedrock AI service, which is designed to simplify the time-consuming process of manual prompt engineering and improve performance by up to 22 percent, depending on the task. The new feature is available acro…
Hacker News — AI stories ≥50 points
TIER_1English(EN)·hackernj·
<p>The post <a href="https://80000hours.org/2026/04/ai-safety-needs-more-than-engineers/">AI safety needs more than engineers</a> appeared first on <a href="https://80000hours.org">80,000 Hours</a>.</p>
<p>While covering Applied Machine Learning Days in Switzerland, Chris met El Mahdi El Mhamdi by chance, and was fascinated with his work doing AI safety research at EPFL. El Mahdi agreed to come on the show to share his research into the vulnerabilities in machine learning that b…
Wewnętrzna presja na wdrażanie sztucznej inteligencji doprowadziła w Amazonie do kuriozalnej sytuacji. Pracownicy, zamiast realnie optymalizować swoją pracę, skupiają się na generowaniu zbędnego ruchu w modelach AI, byle tylko wypełnić korporacyjne wskaźniki KPI i uniknąć negatyw…
Pracownicy Amazona sztucznie pompują użycie AI. Witajcie w absurdalnej erze „tokenmaxxingu” Kiedy wielkie korporacje wydają setki miliardów dolarów na rozwój sztucznej inteligencji, oczekują w zamian cyfrowej rewolucji i skokowego wzrostu wydajności. Rzeczywistość bywa jednak duż…
Amazon employees admit to using AI unnecessarily to pump up internal usage scores — workers complain of intense pressure to use AI tools Amazon is the latest hyperscaler where employees have been caught inflating AI token consumption to hit internal usage targets. https://www. to…
Because using AI tools is definitely a choice for some people. Amazon employees are "tokenmaxxing" due to pressure to use AI tools https:// arstechnica.com/ai/2026/05/ama zon-employees-are-tokenmaxxing-due-to-pressure-to-use-ai-tools/ # Amazon # TokenMaxxing # AI # Labor # Tech
Amazon staff use AI tool for unnecessary tasks to inflate usage scores In-house MeshClaw tool enables employees to delegate jobs to AI agents and climb company’s AI leaderboard Employees at # Amazon , # Meta , and # Microsoft admit to using # AI unnecessarily to pump up internal …
📰 Amazon Employees Are 'Tokenmaxxing' Due To Pressure To Use AI Tools An anonymous reader quotes a report from the Financial Times (via Ars Technica): Amazon employees are using an internal AI tool to automate non-essential tasks in a bid to show managers they are us... 📰 Source:…
Amazon employees admit to using AI unnecessarily to pump up internal usage scores — workers complain of intense pressure to use AI tools Employees at Amazon, Meta, and Microslop have been gaming AI usage metrics. https://www. tomshardware.com/tech-industry /big-tech/big-tech-has-…
Amazon staff are reportedly using AI tools for unnecessary tasks just to boost internal usage scores, a sharp reminder that AI adoption metrics can be gamed. - https:// news.google.com/rss/articles/C BMijAFBVV95cUxQeWNCaHJUTlpVeXM5N09ob3pDNkszaGRoaERoUDhwZkF4SGV3NnJoLVNLNVYxWXNlN…
Wow, "tokenmaxxing" is the new buzzword for # Amazon employees desperately pretending to know # AI 🧠💼—because who needs actual # skills when you can just fake it till you make it! 🙄 AI tools: helping tech workers maintain the # illusion of # competence since yesterday. 😂 https://…
Amazon employees are "tokenmaxxing" due to pressure to use AI tools https:// arstechnica.com/ai/2026/05/ama zon-employees-are-tokenmaxxing-due-to-pressure-to-use-ai-tools/ # HackerNews # Amazon # Employees # AI # Tools # Tokenmaxxing # Pressure # Work # Culture
📰 Amazon AI Quotas Backfire in 2026: Employees Game System with Creative Workarounds Amazon employees are reportedly gaming the system after being forced to meet quotas for AI tool usage. The mandate has led to creative, non-work applications of artificial intelligence, highlight…
📰 2026'da Amazon AI Kotası: İş Güvenliği Tehdit Altında Amazon çalışanlarına yapay zeka kullanım kotası dayatıldığı ve bu kotaların iş güvenliğini tehdit ettiği ortaya çıktı. California eyaleti Amazon'a kota politikaları nedeniyle 5.9 milyon dolar ceza keserek duruma resmi bir bo…
📰 AI Financial Advisor 2026: OpenAI ChatGPT Links to Bank Accounts OpenAI has launched a new feature allowing ChatGPT Pro users in the USA to connect their bank accounts via Plaid for personalized financial analysis based on real transaction data. This marks a significant step to…
📰 OpenAI ChatGPT 2026'de Finans Asistanı Oldu: Banka Hesabını Bağla, AI ile Yönet OpenAI, ChatGPT'nin yeteneklerini finansal alana taşıyor. Artık kullanıcılar, kişisel banka hesaplarını ChatGPT'ye bağlayarak, AI destekli finansal analiz ve yönetim imkanına sahip olacaklar. Bu gel…
📰 AI Usage Mandates Force Amazon Workers to Fabricate Tasks: 2026 Report Amazon workers are creating extraneous and fabricated tasks to meet aggressive new internal mandates for artificial intelligence tool usage. This practice, driven by intense workplace pressure, reveals a sig…
📰 2026'da AI Baskısı: Amazon Çalışanları Sahte Görevler Üretmeye Zorlanıyor Amazon'da yapay zeka kullanımını artırma baskısı altındaki çalışanlar, performans ölçütlerini karşılamak için gerçekte var olmayan görevler üretiyor. Bu durum, şirketin teknoloji odaklı kültürünün insan f…
Some Amazon employees allegedly inflate AI token usage to meet internal weekly AI targets — using an in-house tool for... unnecessary tasks. When KPIs meet AI quotas, creativity finds a way. It's a fascinating reminder that incentive design shapes behavior as much as any security…
Amerykański Departament Handlu rozszerza program audytów bezpieczeństwa sztucznej inteligencji. Do grona firm współpracujących z CAISI dołączyły Google DeepMind, Microsoft oraz xAI, co pozwoli rządowym ekspertom na testowanie modeli przed ich rynkowym debiutem. Inicjatywa ta ma n…
