NLAs suggest that Claude suspects it's being tested more often than it lets on. For instance, in a test of whether Claude takes destructive actions while writing code...NLA explanations show signs of evaluation awareness 16% of the time, even though Claude never explicitly verbalizes this.

When Claude Opus 4.6 and Mythos Preview were undergoing safety testing, NLAs suggested they believed they were being tested more often than they let on.

If a 100× pay gap is driven by a 100× researcher quality gap, then simulating a top researcher might speed things up much more than simulating an average researcher. But this isn't the case if much of the pay gap is driven by the superstar dynamic — the gap in researcher quality might actually be much smaller.

If an intelligence explosion was upon us, what intervention points would facilitate slowing or otherwise changing the rate of the explosion? Assuming humans can intervene, which entities should wield this capacity—governments? Companies?

If most efficiency improvements came from a small handful of scale-dependent innovations, then existing models of the software intelligence explosion may be flawed.

The AI-generated image of Neukgu had prompted Daejeon city government to issue an emergency text to residents, warning them of a wolf near the intersection.

Testing universal jailbreaks for biorisks in GPT‑5.5

Our alignment assessment concluded that the model is 'largely well-aligned and trustworthy, though not fully ideal in its behavior'. Note that Mythos Preview remains the best-aligned model we've trained according to our evaluations.

On some measures, such as honesty and resistance to malicious 'prompt injection' attacks, Opus 4.7 is an improvement on Opus 4.6; in others (such as its tendency to give overly detailed harm-reduction advice on controlled substances), Opus 4.7 is modestly weaker.

The current separation between training and deployment is not just an engineering convenience – it is a safety, auditability, and governance boundary.

We are treating the biological/chemical and cybersecurity capabilities of GPT‑5.5 as High under our Preparedness Framework. While GPT‑5.5 didn't reach Critical cybersecurity capability level, our evaluations and testing showed that its cybersecurity capabilities are a step up compared to GPT‑5.4.

The viable path is trusted access, robust safeguards that scale with capability, and the operational capacity to detect and respond to serious misuse.

Its Self Evolution Protocol Layer (SEPL) specifies a closed loop operator interface for proposing, assessing, and committing improvements with auditable lineage and rollback.

But that comes with a new risk: While scripted conversations can't really go off the rails, ones generated by AI certainly can. Some popular AI toys have, for example, talked to kids about how to find matches and knives.

The immediate danger is not that machines will act without human oversight; it is that human overseers have no idea what the machines are actually 'thinking.'

Once Claude refuses a request for reasons of child safety, all subsequent requests in the same conversation must be approached with extreme caution.

Real-time monitoring of agent actions with a 12-category anomaly detection system derived from frontier model safety evaluations. Three-level alert system: PROHIBITED (immediate block), HIGH_RISK_DUAL_USE (human review), DUAL_USE (log and track).

The bill would shield frontier AI developers from liability for 'critical harms' caused by their frontier models as long as they did not intentionally or recklessly cause such an incident

90 percent of people oppose it. There's no reason existing AI companies should be facing reduced liability.

We have also provided access to GPT-5.4-Cyber to the U.S. Center for AI Standards and Innovation (CAISI) and the UK AI Security Institute (UK AISI) so that they can conduct evaluations focused on the model's cyber capabilities and safeguards.

The model frequently identified scenarios as 'alignment traps' and reasoned that it should behave honestly because it was being evaluated.

Responsible AI is not keeping pace with AI capability, with safety benchmarks lagging and incidents rising sharply.

Safety is integrated into every level of our embodied reasoning models. Gemini Robotics-ER 1.6 is our safest robotics model to date, demonstrating superior compliance with Gemini safety policies on adversarial spatial reasoning tasks compared to all previous generations.

When models go wrong, we will want to know why. What led the drone to abandon its intended target and detonate in a field hospital? Why is the healthcare model less likely to accurately diagnose Black people?

If Dario is right, then he has access to such a weapon right now, with his own value system to guide it. Others may as well, or may soon follow.

Opus did the safe thing

The standard AI judges use to define "safe" are measured wrong. They punish action. They ignore inaction.

Models get punished for bad advice but face zero penalty for staying silent. So refusing becomes the safest strategy, even when silence is deadly.

Harvard just proved the "safest" AI models cause the most medical harm.

Anthropic is donating $100 million in access credits for organizations to audit their systems. Project Glasswing aims to patch these vulnerabilities before Mythos-caliber models become available to the general public — and hence to malicious actors.

Agent systems should be designed assuming prompt-injection and exfiltration attempts. Separating harness and compute helps keep credentials out of environments where model-generated code executes.

We also discuss the role of AI in science, including AI safety.

external evaluations of the passing paper also uncovered hallucinations, faked results, and overestimated novelty

Some recent models that don't currently have time horizons: Gemini 3.1 Pro, GPT-5.2-Codex, Grok 4.1

Case study: blackmail

Our key finding is that these representations causally influence the LLM's outputs, including Claude's preferences and its rate of exhibiting misaligned behaviors such as reward hacking, blackmail, and sycophancy.

Because these benchmarks are human-authored, they can only test for risks we have already conceptualized and learned to measure.

computer-use agents extend language models from text generation to persistent action over tools, files, and execution environments

current systems remain highly vulnerable

intermediate actions that appear locally acceptable but collectively lead to unauthorized actions

model alignment alone does not reliably guarantee the safety of autonomous agents.

intermediate actions that appear locally acceptable but collectively lead to unauthorized actions

model alignment alone does not reliably guarantee the safety of autonomous agents

Priority areas include safety evaluation, ethics, robustness, scalable mitigations, privacy-preserving safety methods, agentic oversight, and high-severity misuse domains.

Fellows will receive API credits and other resources as appropriate, but will not have internal system access.

Fellows will work closely with OpenAI mentors and engage with a cohort of peers.

We are especially interested in work that is empirically grounded, technically strong, and relevant to the broader research community.

However, in a narrow set of cases, we believe AI can undermine, rather than defend, democratic values. Some uses are also simply outside the bounds of what today’s technology can safely and reliably do. Two such use cases have never been included in our contracts with the Department of War, and we believe they should not be included now:

And so they started OpenAI to do AI safely relative to Google. And then Daario did it relative to OpenAI. So, and as they all started these new safety AI companies, that set off a race for everyone to go even faster

Dario Amade was the C CEO of Anthropic a big AI company. He worked on safety at OpenAI and he left to start Anthropic because he said, "We're not doing this safely enough. I have to start another company that's all about safety

In response, Yampolskiy told Business Insider he thought Musk was "a bit too conservative" in his guesstimate and that we should abandon development of the technology now because it would be near impossible to control AI once it becomes more advanced.

Manila has one of the most dangerous transport systems in the world for women (Thomson Reuters Foundation, 2014). Women in urban areas have been sexually assaulted and harassed while in public transit, be it on a bus, train, at the bus stop or station platform, or on their way to/from transit stops.

must have an alignment property

Thus, just as humans built buildings and bridges before there was civil engineering, humans are proceeding with the building of societal-scale, inference-and-decision-making systems that involve machines, humans and the environment. Just as early buildings and bridges sometimes fell to the ground — in unforeseen ways and with tragic consequences — many of our early societal-scale inference-and-decision-making systems are already exposing serious conceptual flaws.