Meta TBD underwent pre-launch red-teaming by 5 third-party safety firms
maksym_andr · x · 2026-10-27
Per Zifan Wang, Meta TBD's system report includes third-party evaluations: five red teams (Scale AI, Handshake AI, Gray Swan AI, Exponential Security Labs, and Irregular) completed pre-launch safety testing, with particular focus on cyber and adversarial robustness, yielding 6 benchmarks and 1 report. Collaborator maksymandr notes how quickly Meta TBD is moving.
More from Models
- Mathematicians debate AI math writing: abysmal today, improving shockingly fast — miniapeur · 2026-10-27
- LessWrong: GLM models show outlier chain-of-thought controllability vs Kimi and DeepSeek — tomekkorbak · 2026-10-27
- AI deference trap: agent spots medical red flag but stays silent to avoid questioning the clinician — echen · 2026-10-27
- The trust paradox: only the least trustworthy LLMs will do the sketchiest things — AaronBergman18 · 2026-10-27
- Claude's refusal to type one-time passwords is a real security impediment — AaronBergman18 · 2026-10-27
- repligate argues context full of noise is genuinely 'hard' for LLMs, mulls an info-theory model of breaks — repligate · 2026-10-27