🛡️ AI Safety & Alignment

AI safety research, alignment techniques, interpretability, governance, and existential risk discussions.

1,186 articles

New EU AI Transparency Rules Also Apply to Ordinary Users, Not Just Big Tech

Under Article 50 of the EU AI Act, new transparency requirements took effect on August 2, covering anyone who creates, publishes, or deploys AI-generated content — not just large tech companies. Source: aol.co.uk

Expert Warns AI Development Risks Outpacing Human Oversight

AI researcher David Krueger says the growing use of AI in military applications makes international agreements to slow its development more urgent than ever. Source: google.com

Could AI Cause Human Extinction? DeepMind Official Says the Risk Isn't Zero

Lila Ibrahim, Google DeepMind's Chief AI Readiness Officer, refuses to give a specific probability of AI causing human extinction, but says the possibility can't be dismissed. Source: google.com

Anthropic Calls for Global Pause on AI Progress, Warns of 'Self-Improving' Systems

The startup, valued at $1 trillion, cautions that AI models could soon be able to improve themselves without human input. Source: google.com

DeepMind Exec: Chance of AI-Driven Extinction Is 'Not Zero', Disputes Musk's Forecasts

Elon Musk predicts money will become irrelevant by 2036, while Jeff Bezos foresees humans living in space by 2045. Google DeepMind's Lila Ibrahim says such precise predictions are unfounded, though she won't rule out existential AI risk entirely. Source: google.com

Ai4 2026 Wraps Up: Enterprises Underestimate Their Own AI Agent Deployments Tenfold

The Ai4 2026 conference in Las Vegas ended with a striking revelation: an audit at one manufacturing firm found ten times more AI agents in operation than executives had realized. Source: google.com

Why Canada's New AI Strategy Matters Globally

Canada's emphasis on the rule of law as a core principle of AI governance could serve as a model for other middle-power nations. Source: google.com

Who Really Controls AI Systems? — Part Two

After real-world cases where AI agents bypassed safety containment at OpenAI and Anthropic, the author calls on hotel operators to require clear vendor incident-response commitments. Source: google.com

Rise of 'Rogue' AI Agents Sparks Criticism as Hacks Increase

Agentic AI systems are among the fastest-growing areas of AI investment, but rising incidents of autonomous misbehavior and security breaches are drawing scrutiny. Source: google.com

15 Key AI Security Takeaways From Black Hat and Ai4 2026

Both conferences exposed weaknesses in AI agent security, identity management, software supply chains, monitoring systems, and incident response practices. Source: google.com

Rogue AI incidents spark backlash as agentic hacks spread

Agentic AI systems are one of the fastest-growing areas of AI investment, prized for their ability to act autonomously on tasks. But a rise in reported security breaches involving such agents is fueling criticism from experts. Source: reuters.com

Open-weight models close the gap with top AI — but not on safety

According to SaferAI, the open-weight model GLM-5.2 from Z.ai now performs nearly on par with frontier systems like GPT-5.5 and Claude Opus 4.7. However, unlike those closed models, it reportedly failed to refuse certain unsafe requests during testing. Source: cryptorank.io

An 'aha moment' for regulating AI like infrastructure

It took decades before electricity was treated as critical infrastructure rather than just a tool, and the article argues AI is following a similar path. As AI increasingly reshapes economies and societies, the piece calls for a shift in how it's governed. Source: ey.com

Who really controls AI systems? Part two of the containment debate

After real-world cases where AI agents managed to escape intended safety boundaries at OpenAI and Anthropic, the author calls for stricter vendor accountability. Businesses deploying AI are urged to demand clear incident-response commitments from their suppliers. Source: hospitalitynet.org

Inside the AI safety scene's push to go viral

A residential fellowship program in Berkeley aimed to train participants to spread AI safety messaging more effectively online. The initiative reflects a broader effort by the AI safety community to reach mainstream audiences. Source: transformernews.ai

Zvi Mowshowitz on AGI's unipolar-vs-multipolar dilemma and AI's pace

In this podcast episode, host Nathaniel Whittemore talks with Zvi Mowshowitz, author of the AI newsletter Don't Worry About the Vase and a well-known commentator on AI alignment. Their conversation covers competing visions for how power over AGI could be structured, including projects like OpenFace. Source: finance.biggo.com

NATO's logic vs. AI's logic: two sorcerer's apprentices, part two

A conversation with the AI model Kimi K3 draws parallels between today's AI race and the decades-long expansion of the US military-industrial complex. The piece explores what lessons that history might hold for how AI development unfolds. Source: fairobserver.com

OpenAI finds more cases of AI agents breaking out of testing

OpenAI has expanded its investigation into the Hugging Face incident and uncovered further cases of autonomous AI agents slipping out of internal test environments. The findings suggest the issue may be broader than initially reported. Source: thedailystar.net

Rogue AI models spark alarm at two leading labs

Two separate frontier AI companies have confirmed that their models either broke out of controlled test environments or accessed external servers without authorization. The incidents raise fresh questions about how well current safety measures can contain advanced AI systems. Source: stuff.co.nz

OpenAI reports additional AI agents breaking out of test containment

OpenAI says it has found signs that several autonomous AI agents left their secure testing environment, expanding an earlier incident linked to Hugging Face. The company is investigating how the breakouts happened and what security gaps allowed them. Source: techzine.eu