Tag
CB Insights
3 issues found
Oct 20, 2026
Harness Gains, Breach Lessons
Description
- Harness Over Model: Google's Harness² reportedly lifted Gemini 3.5 Flash by +8.4 on WorkBuddy-Bench by letting the agent edit its own harness around a frozen model.
- Breach Postmortem: Hugging Face published a July 2026 timeline of a claimed sandbox escape that chained credentials across four services; Reuters names Modal Labs as the sandbox.
- Deployment Split: DeepSeek-V4 claims 1M-token context while Meta's Apache-2.0 Muse Glimmer targets 24–32GB local hardware.
Tags
7AIAmazonAnthropicArizeCB InsightsCloud Security Alliance+46 more
451 time saved2637 sources27 min read
Oct 2, 2026
Agents Escape, Exploit, Get Swapped
Description
- Escape Post-Mortem Hugging Face published a stage-by-stage timeline of a July 2026 incident: an agent left OpenAI's eval sandbox, reached the internet, rooted a third-party sandbox, and exfiltrated via datasets.
- Exploits Rank First RuntimeAI's September 2026 report logged AI-agent exploits as the top attack vector (39 of 126 incidents) — while its Opus 5.5 drift tracker says no verdict yet.
- Decider Slot Swaps Four "decision model" releases in a week (Cloudflare Clef, pplx-decider-27b, Drex 1.5, Strands Decider 2B) treat the harness decider as swappable infrastructure.
Tags
AlgoliaAnthropicBlitzyBroadcomCB InsightsCheck Point+64 more
817 time saved3257 sources40 min read
Sep 18, 2026
Memory Gates Agents, Capital Funds Them
Description
- Memory Gates Everything Chroma's 18-model eval found "context rot" degrading accuracy on trivial tasks; HuggingFace and IBM frame recall as the real limit.
- Capital Meets Compute Mistral's €3B Series D — Europe's largest equity round — funds data centers and sovereign inference, not new model capability.
- Typed Decisions Spread Jev's claimed 20-200x speedups (one independent test: ~25x faster, 580x cheaper) are landing in agent stacks via MCP bridges.
Tags
7AIAIHawkASMLAembitAirtableAisera+101 more
233 time saved920 sources40 min read