Latest
  1. Johns Hopkins Bayview tells 3,182 patients an employee put their health data into a public AI tool
  2. China’s AI compute access expected to double in 2027, with US-designed chips about 65 per cent
  3. UK commits £100m to national AI Biosecurity Lab
  4. Moonshot AI releases Kimi K3.1 Pro with 1-million-token context window
  5. Residents in Newport and Cardiff oppose hyperscale datacentres near homes
  6. Study: Grok 4.7 picked Republicans in 92% of forced midterm answers, GPT-6.1 Sol picked Democrats in 97%
  7. Attackers exploiting the PHP flaws hand the break-in to AI agents, Unit 42 says
  8. Elvis Presley’s rightsholders sue Anthropic over training on 274 songs

Latest / Frontier models and safety

ReportFrontierHealthUSConfirmed

xAI publishes biosecurity results for Grok 4.6 from an outside evaluator

See how this story connects

Source: xAI

xAI published its approach to biosecurity alongside an independent analysis of Grok 4.6 by LatchBio. On BioSecBench-Refusal, Grok 4.6 refused 59.2% of red-team tasks and completed 64.8% of routine ones, the only tested model above 50% on both, averaging 62.1% on the combined metric; on BioSecBench-Surveillance it averaged 53.5%, behind Opus 5 and ahead of GPT-5.6 Sol. xAI describes refusal training, inference-time filtering, deployment controls and session- and user-level monitoring, says it tests biological capability before release, and says it treats over-refusal of routine biology as a risk equal to aiding misuse. It plans broader pre-deployment suites, more third-party evaluations and deployments with biology institutions.

Why it matters

xAI is now publishing third-party biosecurity results for its model.

When
1 Sep 2026
ID
WTA-N-2026-0259
Who
xAI, LatchBio

Summary by wheresthe.ai, from xAI. Checked against the source on 12 Oct 2026.

Read the original at xAI

Source copy: download the page as saved on 18 Oct 2026 (PDF). Document the source offers: document 1 (PDF). Also held by the Internet Archive.

Next stop on this Storyline: What happened nextAnthropic opens a verification programme giving vetted biologists looser safeguards

Cite this story

WTA-N-2026-0259: “xAI publishes biosecurity results for Grok 4.6 from an outside evaluator.” wheresthe.ai, 1 Sep 2026. https://wheresthe.ai/d/xai-biosecurity-at-the-frontier-2026-09/ (source: xAI).

Spotted an error? Email [email protected].