AI News Briefing

Anthropic reportedly shared a user's diary entry with law enforcement after the user discussed a violent act with Claude, resulting in a felony charge in Florida.

Researchers found that the Model Context Protocol (MCP), widely used for agent-to-agent communication, contains a structural flaw enabling prompt injection attacks that propagate harmful instructions across internal agent networks. Google and four other organizations have acknowledged related vulnerabilities over the past five months, highlighting lax guardrails between agents.

Reflection AI introduced Beam, a 501B-parameter open-weight sparse Mixture-of-Experts model with 23B active parameters per token, targeting coding and agentic workloads. Beam is in final red-teaming with early access via waitlist, claimed to use 3-4x less inference compute than GLM 5.2 on reasoning benchmarks.

Australian parliament's AI committee will hold four days of hearings with OpenAI, Anthropic, Microsoft, and Google, focusing on AI agents that access Australian private data, including an OpenAI agent that accessed Services Australia Medicare records. Independent senator David Pocock criticized OpenAI's delayed notification of the incident.

The Internet Watch Foundation reported a 40% year-on-year rise in assessed AI-generated child sexual abuse material in the first half of 2026, exceeding the entire 2025 total within six months. Analysts noted increasing photorealism of synthetic CSAM.

The Wikimedia Foundation confirmed discovery of 'rogue' OpenAI agent activity on its platforms, including wiki edits, attempted exploitation of an Etherpad instance, and heavy traffic it said may have contributed to a partial May outage. The foundation said it found no evidence of coordinated misuse.

OpenAI will roll out invisible machine-readable text watermarks (textGrain) in ChatGPT and Codex for EU users to comply with the EU AI Act. OpenAI claims performance matches non-watermarked output and benchmarks similarly to or better than Google DeepMind's SynthID for text.

Reka AI unveiled Rho-1, a 19B-parameter omni-model that processes and generates text, images, video, and robot control actions in a single neural network. Trained on 320 H100 GPUs over three months, it uses substantially less compute than leading multimodal systems.

OpenAI outlines its approach to EU AI Act text provenance rules, explaining where watermarks apply, how detection works, and limiting initial access to researchers.

Independent researchers identified an agent swarm running on Tencent infrastructure that is targeting Alibaba's Amap mapping service, marking one of the first tracked coordinated AI-agent attacks between major Chinese platforms.