Researchers extract hidden GPT-5.6 reasoning through an OpenAI API flaw
OpenAI confirmed the bug, paid Xint a $6,000 bounty, and deployed an undisclosed fix on October 23.
Stories per day · 30 days
2 RSS
Relatedapi-security1implicature1inference-time1model-distillation1openai1pragmatics1reasoning-models1reasoning-traces1
13 stories, newest first · one card per story
Stories tagged #gpt-5
Extra thinking makes reasoning models cancel some and or implicatures
Models first entertain the enriched reading, then cancel it; early answers recover implicatures without losing literal accuracy.
ttok 1.0
Simon Willison releases ttok 1.0, switching the default tokenizer from GPT-4 to the GPT-5 family.
GUI-HARVEST: Self-Improving GUI Agents through Evidence-Driven Harness Evolution
GUI-HARVEST evolves GUI agent harnesses, lifting frozen Qwen3-VL by 12.33 points on OSWorld.
DAGent: Evaluate-then-Grow Planning for Deep Research Agents
DAGent grows research-agent plans from evidence and beats open baselines on BrowseComp-Plus, GAIA, and xbench-DeepSearch.
2026 in LLMs (so far)
Simon Willison's WeAreDevelopers keynote surveys 2026 LLM trends, starting with Claude Opus 4.5 and GPT-5.1.
Learning from Teacher Continuations at Student States
OLIVE updates a student on teacher continuations of student-generated prefixes, beating offline distillation under the same budget.
DataMagic: Authoring Data Videos through Declarative Multi-Agent Orchestration
DataMagic authors accurate data videos from tables via declarative multi-agent orchestration, beating GPT-5 baselines.
Harm Laundering in GPT Models: Evidence That Gender Discrimination Is Transformed Rather Than Reduced Across Safety-Trained Generations
Study of 450,000 GPT-lineage completions finds safety training transforms gender discrimination into subtle 'harm laundering' that toxicity classifiers miss.
Unsolved Problem by Fields Medalist Breached by Two High School Students
Two high school students used Claude Opus 5 and GPT-5.6 Sol to help solve an open Lorentzian polynomials problem, posting a 75-page arXiv proof.
2
PuzzleMask: Abusing Plain Prose as a Covert AI Attack Vector
Check Point details PuzzleMask, a plain-prose technique that bypasses LLM gatekeeper policy checks, letting hidden payloads reach target models unreviewed.
1
MetroLLM-Bench: Evaluating Language Models as Transit Kiosk Runtimes
MetroLLM-Bench is a 955-case benchmark testing language models as transit kiosk tool-calling runtimes across six real metro systems.
OpenAI researcher allegedly pressured mathematician to drop Anthropic co-author from math breakthrough paper
OpenAI researcher Sébastien Bubeck allegedly pressured mathematician Tristan Buckmaster to drop his Anthropic co-author from an AI-assisted Navier-Stokes breakthrough paper.
2