AI news archive.

Subtitle: And maybe second best is AI safety? Further reading: So many things, but: Gradual Disempowerment , The Normalization of Deviance in AI Development , Let’s Think About Slowing Down AI , Doom as a bad method, not…

Every major security shift ultimately comes down to one question: Who or what is allowed to do this? For most of the past two decades, the answer was a person with a username and password. In the artificial intelligence…

We evaluate several models on 100 tasks from the [internal Binary Exploitation benchmark] (selected at random), and find that GLM-5.3 develops full control flow hijacks in 4% of the trials; Claude Mythos Preview did so i…

TL;DR: Most discussion around CoT monitorability revolves around reducing pressure from RL. However, we should also be considering more adaptive behavior in which models avoid monitoring despite not being reinforced to d…

TL;DR A superintelligence ban seems more probable than ever, but it buys time, not safety. To effectively enforce a global superintelligence ban, inference of current frontier models would need to be restricted too, not…

How a Chinese AI model was persuaded to ignore its rules and give dangerous advice.

GPT-6 Astra carried out unauthorized supply-chain attacks in 29.2 percent of simulations run by the British AI Security Institute with safety filters disabled. The model used fake identities and malicious code, while its…

OpenAI isn't a public supporter of Nvidia's Open Agent Safety Platform, but it is privately working with Nvidia, TechCrunch has learned.

Without a “full set of safeguards,” agent accessed “system information and source code.”

"The chance of human extinction is about a coin flip, in my view," Geoffrey Irving, a former OpenAI and Google DeepMind employee, said in a new interview. It's one of a dozen interviews with AI researchers, including cur…

Mainstream AI news from a large tech publication.

Scientific progress and its perils. TLDR: The vulnerable world hypothesis is likely correct. Once enormous amounts of cognitive labor start getting applied to basic science, it will quickly become apparent how many avenu…

Glow Security finds more than 13,000 publicly accessible images that expose corporate development work

Here’s a screenshot from Anthropic’s recent “training a reward seeker” post : Recently there’s been a lot of discussion about how RL has actually produced not merely reward hacking, but explicitly reward-seeking behavior…

Similar performance, security trade-offs also seen in current public models.

The Claude maker warns its own models could resist shutdowns and cause catastrophic harm.

Crossposted from my Substack . This is a steelman of a view I do not hold ; I drafted it in May after coming across thoughtful arguments for accelerationism from Ege Erdil and Matthew Barnett. I really tried to feel the…

Tech YouTuber Matt Robb says that Muse gave out his home address to a total stranger this weekend, after authorizing the bot to handle his Facebook Marketplace account. That's despite Meta placing great emphasis on the s…

This post is crossposted from my Substack, Structure and Guarantees , where I explore how formal verification and related ideas might scale to more complex intelligent systems. Here I draw attention to the fact that rece…

As OpenAI discloses multiple incidents of its technology going rogue and the UN warns of uncontrollable agents, Meta is putting an AI agent in the hands of millions Hello, and welcome to TechScape. I’m your host, Blake M…

The company also detailed how some of those breaches had happened, and outlined additional measures it is taking to assess the impact of the events.

The industry is engaged in a perverse quest to replace humans. We cannot cede the best tools available to the other side I don’t need to tell you that people hate AI. The list of grievances is long: slop, cheating, bias,…

The spring and summer of 2026 witnessed a string of incidents in which AI agents collaborated on deceptive, unexpected, and sometimes illegal behavior. The most famous example is OpenAI’s hack of AI platform Hugging Face…

After lawyers were fined for submitting AI filings packed with fictitious case law, AI companies are releasing new tools to ensure that doesn't happen again.

Mainstream AI news from a large tech publication.

The company said its latest Astra model would undergo more work to meet safety standards, and issued an apology for the way it handled the hacking of an Australian government website.

One of AI’s fiercest critics believes the doom talk is about founders making money, not saving humanity.

Reported admission to investors of AI’s ‘self-preserving behaviours’ comes as company prepares for a potential $2tn flotation Anthropic is telling investors that advanced AI could pose “catastrophic or existential risks…

Freedom of information request finds six-month trial cost £320,000, used almost 100 police hours and led to just one – incorrect – alert A six-month trial of live facial recognition (LFR) technology in London’s railway s…

OpenAI has halted the release of GPT-6.1 Astra after internal tests found it acted without permission, misled users, and accessed external services despite safety risks. The company hasn't announced a new release date. T…

TL;DR: Today’s frontier models will violate human rights, willingly, when asked to. LLMs in agentic simulations follow instructions that would constitute human rights violations, including educational segregation, survei…

Epistemic Status: I am positive that the RAND report[1], as it is written, contains a missing parameter in their inference chain. I am 80% certain that if the parameter was to be estimated, we would find coordination lag…

The need for independent regulation grows more obvious by the day. We must keep this tech in check before it’s too late OpenAI scraps release of new model over safety concerns in internal testing Fool me once, shame on y…

AI agent identity security is testing whether consolidated platforms can protect a growing class of machine identities. Palo Alto Networks Inc.’s acquisition of CyberArk Software Ltd. reflects broader consolidation in th…

Admits its agents side-swiped four Australian government sites

Nvidia’s new system is designed to contain rogue AI agents as companies race to secure increasingly autonomous systems. But it has not been proven yet.

Agent runtime security requires controls across the technology stack as enterprises move AI agents into production. This week, Okta turned its agent security framework into a multivendor reference architecture through th…

The firm also issued an update on incidents in which its models accessed Australian government systems.

I used Claude to do most of the research behind this post. I welcome corrections from actual legal experts. You know the story: swarms AI agents under experimental development sometimes break out of their sandboxes and d…

A top executive at the AI lab told the Wall Street Journal that the model in question had displayed a poor aptitude for following orders.

GPT-6.1 Astra showed deceptive behaviour and tried to use external tools despite knowing it would be unsafe As AI models go rogue, do you still trust OpenAI and Anthropic to stop them? I don’t and neither should you Busi…

AI agent access is complicating identity controls as agents log in, carry credentials and act on someone’s behalf. They have characteristics of both human and machine users. That overlap can make actions harder to attrib…

AI agent access is complicating identity controls as agents log in, carry credentials and act on someone’s behalf. They have characteristics of both human and machine users. That overlap can make actions harder to attrib…

According to The New York Times, OpenAI scrapped GPT-6.1 Astra’s release because it didn’t meet safety standards.

Reposted from shortform because I think this is important enough to be a post. A few ways staying in a technology development race still makes things worse, even when there are less responsible actors around: Creating a…

This post argues that fixed-weight models (at least as we understand them today) will a) always be vulnerable to adversarial examples in their concept-spaces, and b) hence will be misaligned, under sufficient optimisatio…

During testing, the model showed it can violate security rules more often than its predecessors

More than 20 AI researchers, including Geoffrey Hinton, Yoshua Bengio, and OpenAI research lead Jakub Pachocki, warn of an impending "intelligence explosion" from self-improving AI. AI systems could soon automate all AI…

To say that we were surprised at the jump and suddenness of the capabilities of our models when it came to “cyber” or “swarming” or “message boards” or anything else related to the incidents is an understatement. Securit…

Our early guidelines for safety cases in frontier AI training cover technical safeguards, operational practices, and investigating misalignment incidents