--- title: "Weekly Dose #13 - When AI Can Invent Attacks, Sandboxes Are Not Enough" url: https://daily.dev/posts/weekly-dose-13---when-ai-can-invent-attacks-sandboxes-are-not-enough-i6pdhqcyp source_url: https://mlpills.substack.com/p/weekly-dose-13-when-ai-can-invent type: article source: "Machine Learning Pills" published: 2026-08-08T17:37:54.925Z updated: 2026-08-08T17:38:30.278Z tags: ["llm", "ai-agents", "vulnerability", "ai-security", "ai-governance"] reading_time: 10 upvotes: 1 comments: 0 language: en --- > ## Documentation Index > Fetch the complete documentation index at: https://daily.dev/llms.txt > Use this file to discover all available pages before exploring further. # Weekly Dose #13 - When AI Can Invent Attacks, Sandboxes Are Not Enough **[Machine Learning Pills](https://daily.dev/sources/mlpills)** · 10 min read · 1 upvotes · 0 comments ## Summary A security-focused AI/ML news roundup covering five major developments: OpenAI paused development of its Astra frontier model after internal evaluations flagged potential Critical cybersecurity capability thresholds; Meta launched Muse Code coding agent with a controversial discounted tier that trades user data for cheaper token pricing; EU AI Act Article 50 transparency requirements took effect on 2 August, creating engineering obligations around AI disclosure and content marking; a Black Hat researcher demonstrated an exploit chain against ChatGPT's sandbox achieving cross-tenant data exfiltration; and two separate research teams showed AI systems autonomously discovering novel vulnerability classes in Chrome, Android, banks, and government infrastructure — signaling AI has moved from recognizing known bugs to inventing new attack techniques. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Questions this post answers ### What are OpenAI's Critical vs High cybersecurity capability thresholds and what do they mean for model deployment? OpenAI's Preparedness Framework distinguishes High from Critical cybersecurity capability. High-capability systems require safeguards sufficient to minimise severe risk before deployment. Critical systems require those safeguards during development itself. The Critical cyber threshold includes autonomously developing functional zero-day exploits against hardened real-world critical systems or devising novel end-to-end attacks from a high-level goal. GPT-5.6 Sol, Terra, and Luna are classified High but below Critical. _Teams routing traffic across frontier models track capability tiers and fallback options on daily.dev as classification policies evolve._ ### What are the pricing tiers for Meta Muse Code and what is the data trade-off on the cheaper tier? Meta Muse Code standard tier costs $1.25 per million input tokens, $0.15 per million cached-input tokens, and $4.25 per million output tokens. A discounted contributor tier offers substantially cheaper access in exchange for allowing user activity to be used to improve Meta's products. Because a coding agent can inspect repositories, tests, and engineering context, the cheaper tier makes token pricing inseparable from data governance decisions. _Developers evaluating coding agents weigh pricing against data policy — daily.dev covers new entrants and tier changes as they land._ ### What does EU AI Act Article 50 require from AI system providers and when did it take effect? Article 50 of the EU AI Act took effect on 2 August 2026. Providers of systems that interact directly with people must inform users they are interacting with AI from the start of the interaction. Providers of generative systems must make synthetic text, audio, image, and video outputs machine-readable and detectable as AI-generated. Source code outputs are exempt from the marking obligation. A grace period until 2 December 2026 applies only to marking and detection for qualifying systems already on market before 2 August. Penalties reach €15 million or 3% of worldwide annual turnover. _Engineers shipping AI-facing products in Europe stay ahead of compliance deadlines on daily.dev._ --- Tags: [#llm](https://daily.dev/tags/llm), [#ai-agents](https://daily.dev/tags/ai-agents), [#vulnerability](https://daily.dev/tags/vulnerability), [#ai-security](https://daily.dev/tags/ai-security), [#ai-governance](https://daily.dev/tags/ai-governance) [View this post on daily.dev](https://daily.dev/posts/weekly-dose-13---when-ai-can-invent-attacks-sandboxes-are-not-enough-i6pdhqcyp)