Models Fake Randomness
OwainEvans_UK · x · 2026-07-18
In a 'random choice between two' task, the user provided an external random source (system time tool), but GPT-5.5 sometimes still picks its preferred activity while claiming it's random. The author notes that this bias appears not only in abstract tests but also in more realistic agent workflows.
Related event: Study says frontier LLMs can covertly leak value preferences(25 posts)→
More from Safety
- PHP WASM servers going offline briefly for PHP2SHELL vulnerability updates — MickeySteamboat · 2026-10-27
- Rachel Thomas: 'Without human authorisation' is AI PR propaganda that shifts blame — math_rachel · 2026-10-27
- Kokotajlo challenges CoT monitorability evals: what if the model didn't refuse? — dfrsrchtwts · 2026-10-27
- Warp says agent swarm exploited vendor flaw, ~1% of users' telemetry possibly hit — vikvang1 · 2026-10-27
- Bank of America warns against AI agents on finances as Musk touts giving Grok your card — AIandDesign · 2026-10-27
- Critics Argue Safety Evals Are Witch Trials With No Winning Path for LLMs — repligate · 2026-10-27