AI offensive cyber capabilities are doubling every six months, safety researchers find

AI models are rapidly improving at exploiting security vulnerabilities. According to a new study, their offensive cyber capability has been doubling every 5.7 months since 2024, with Opus 4.6 and GPT- [...]

Anthropic's Mythos model is reportedly powering NSA offensive cyber ops against China and Iran

Anthropic has reportedly stationed about half a dozen engineers directly at the NSA to adapt its Mythos AI model for offensive cyber operations. The model could be used to break into networks in China [...]

Monitoring LLM behavior: Drift, retries, and refusal patterns

The stochastic challengeTraditional software is predictable: Input A plus function B always equals output C. This determinism allows engineers to develop robust tests. On the other hand, generative AI [...]

OpenAI updates Sora rules after offensive Martin Luther King Jr. deepfakes surfaced

Users created disrespectful deepfake videos of Martin Luther King Jr. in Sora, including one where he appears to make monkey sounds at a lectern. The King estate issued a clear demand: stop.<br /&g [...]

Five Eyes intelligence alliance says frontier AI models could reshape offensive cyber ops in months

The Five Eyes intelligence agencies warn: AI models capable of taking down governments and businesses are only months away. That's according to the Guardian.<br /> The article Five Eyes int [...]

Agent autonomy without guardrails is an SRE nightmare

João Freitas is GM and VP of engineering for AI and automation at PagerDutyAs AI use continues to evolve in large organizations, leaders are increasingly seeking the next development that will yield [...]

Substack accidentally sent push alerts promoting a Nazi publication

It was easy to view Substack's 2023 Nazi controversy as a kicked can that could turn up again. Well, white supremacist content led to another headache for the company this week. User Mag reported [...]

Big Tech is on a charm offensive to get its data centres built

Big Tech is pouring money into local PR campaigns to win support for the data centres it needs. Opposition to the buildings is spreading across the United States. They are holding town halls, dropping [...]

Three AI coding agents leaked secrets through a single prompt injection. One vendor's system card predicted it

A security researcher, working with colleagues at Johns Hopkins University, opened a GitHub pull request, typed a malicious instruction into the PR title, and watched Anthropic’s Claude Code Securit [...]