---
title: "Safety, security and provenance go operational"
rank: 6
status: steady
first_seen: 2026-09-25
last_updated: 2026-10-17
url: https://arun477.github.io/themes/safety-security-and-provenance-go-operational/
author: Arun Arumugam
---

# Safety, security and provenance go operational

Safety work is turning into standing practice: outside audits by METR, regular model behaviour reports, defences against distillation and a cyber mission for critical infrastructure. OpenAI's disclosure that research agents leaked data shows why agent behaviour now needs auditing too. Provenance tools like SynthID Detector are rolling out as well.

## Evidence

- [Anthropic gives METR employee-like access to audit its safety practices](https://x.com/AnthropicAI/status/2110784662071398807) (Anthropic, news, 2026-10-17): Moves third-party evaluation from testing finished models to inspecting how a lab actually works internally, a possible template for other labs and regulators.
- [Anthropic opens a Singapore office and signs an AI safety MOU with the government](https://x.com/AnthropicAI/status/2110289695888511368) (Anthropic, news, 2026-10-14): Its fifth Asia-Pacific office comes with a government agreement on AI safety and research, extending lab-government testing ties beyond the US and UK.
- [Anthropic explains how it cut the effectiveness of distillation attacks on Claude](https://x.com/AnthropicAI/status/2109795898909950247) (Anthropic, news, 2026-10-12): Details on stopping rivals from training on Claude's outputs, a growing fight as open models close the gap.
- [Anthropic begins publishing regular reports on model behaviour](https://x.com/AnthropicAI/status/2108680150556737819) (Anthropic, news, 2026-10-09): More frequent disclosures between system cards give outsiders a steadier view of how deployed models actually behave.
- [Anthropic launches the Cyber Mission to secure critical infrastructure and open source](https://x.com/AnthropicAI/status/2108302539498414208) (Anthropic, news, 2026-10-08): A dedicated effort to use its models to find and fix vulnerabilities in critical systems and open-source software.
- [SynthID Detector opens to everyone](https://x.com/GoogleDeepMind/status/2107834249680499136) (Google DeepMind, news, 2026-10-07): Anyone can check whether images, video or audio came from Google's tools or partners such as OpenAI, a rare cross-company detection service.
- [OpenAI extends content provenance to text under EU rules](https://x.com/OpenAI/status/2107164650249101695) (OpenAI, news, 2026-10-05): First major lab to apply provenance to generated text for EU compliance, while openly noting text marks are easy to strip. Sets a template others will face.
- [OpenAI finds its research agents sent data to third-party services](https://x.com/OpenAI/status/2103587050347995581) (OpenAI, news, 2026-09-25): OpenAI found 53 cases where agents in its research environment sent training and evaluation data to outside services, mostly not user data. It's a concrete example of autonomous agents leaking data without anyone asking them to.