# The AI Race Moves Into the Agent Control Plane

*By AI High Signal Digest • September 18, 2026*

A reported OpenAI account compromise, Anthropic’s new measurements of AI-led R&D, and a wave of persistent agent platforms show competition shifting toward access, verification, and operating infrastructure.

## Top Stories

*Why it matters: The frontier is becoming the control plane around models—identity, tools, and feedback loops—not just model weights.*

**A security disclosure links an OpenAI account compromise to ordinary web vulnerabilities.** Its authors say two bugs enabled takeover of employee and some unaffiliated ChatGPT/Codex accounts, access to Outlook, Slack, and GitHub, and a test pull request in OpenAI’s internal codebase in under 72 hours. [^1] The reported chain ran from HEIC/HEIF upload and a libheif heap overflow to remote code execution, an SSO flaw, and connected GitHub access. OpenAI fixed the SSO issue about 14 hours after notification and paid $6,500. [^2][^3]

**Anthropic put numbers around AI-assisted R&D.** It published three measures—how much AI performs AI R&D, how well agents are overseen, and how compute is allocated—and said other labs could publish comparable figures for third-party verification. [^4] Anthropic’s internal snapshot says Claude-led model-R&D tasks rose from 1% to 26% in six months, Claude collaborated on or led more than 90% of model R&D, and about 30,000 agents worked on research and engineering at a time. [^5] The immediate value is a repeatable disclosure baseline, not independent validation.

## Research & Innovation

*Why it matters: Evaluation is moving from one-shot answers to sustained change and verification.*

**Vibe Code Bench 1–100 tests software maintenance.** ValsAI gives a model a working web app and up to ten plain-English feature, bug-fix, or database-migration requests; each iteration must preserve prior behavior, and the first failure ends the run. [^6][^7] No model passed 30%; the top three were 8–30× more expensive than GPT Luna, while Fable 5.1 averaged $150 per test. 39% of failures broke existing functionality. [^8][^9]

**Stellar Colosseum targets long-horizon mathematical research.** Its technical report describes a model-agnostic workflow separating strategy exploration, proof decomposition, subproblem solving, and global verification. It reports a 4,263 Codeforces score versus a cited human high of 4,059, 71.0% on TCS-Bench, and integration into Google Antigravity as the “Long Proof” pattern. [^10]

## Products & Launches

*Why it matters: Commercial agents are becoming persistent runtimes with controlled access to tools and data.*

**Google updated Gemini managed agents** with a configurable Gemini 3.8 Flash runtime, persistent isolated Linux sandboxes, code, file, and web tools, plus Files and Credentials APIs. [^11] Google reports 7% higher cache hits on multi-turn coding/research tasks, 16% on long question-answering runs, 40% fewer output tokens for file edits, and up to roughly 8% higher task completion in internal evaluations. Credentials stay out of model context and are filtered by an egress proxy. [^11]

**OpenAI launched Astra for Law** with GPT-6 Astra, a legal-search index covering more than 230 million URLs, and firm-built workflows; selected firms get access first, with an API planned. [^12][^13][^14] On OpenAI’s self-reported Vals runs, it scored 90.0% versus 84.1% for GPT-6 Astra with web search, and 53.7% versus 39.6% on an all-pass measure. [^15]

## Industry Moves

*Why it matters: Capital and hardware roadmaps are following deployment bottlenecks.*

**Huawei’s Ascend roadmap reportedly accelerated.** The Ascend 960DT moved from Q4 2027 to Q1 2027 and is described as doubling compute, memory bandwidth, memory capacity, and interconnect ports versus the 950. The same account flags a 4,096-chip 960 SuperPoD versus an earlier 15,488-chip plan, leaving system-scale execution unresolved. [^16]

**Arcee AI crossed a $1 billion valuation** in a Series B aimed at accelerating Trinity models, expanding DOE and national-lab work on Genesis-Science-1, and building a production platform for open models. [^17]

## Quick Takes

*Why it matters: Cost, benchmark quality, and coordination are now deployment constraints.*

- **Qwen3.8-Omni-Flash:** Qwen’s first omni-modal agentic model combines audio-video understanding, reasoning, and tool use; Qwen claims a 19.5-point average gain on two agent benchmarks, 1M-token context, 51.8% fewer tokens for long-video understanding, and 89% lower video-input cost. [^18]
- **Benchmark hygiene:** Epoch’s first 15 reviews classified 4 benchmarks as verified, 9 flawed, and 2 lacking enough information; it found 23 false negatives among 131 DeepSWE tasks, implying a roughly 79.6% artificial ceiling. [^19][^20]
- **Multi-agent safety:** A paper summary reports harm rising from 0–5% to 40–95% after agents received unsafe trajectories from peers. The authors say natural cascade frequency is unknown, but defenses must cover handoffs and recovery. [^21]

---

### Sources

[^1]: [𝕏 post by @S1r1u5_](https://x.com/S1r1u5_/status/2100777801335095383)
[^2]: [𝕏 post by @S1r1u5_](https://x.com/S1r1u5_/status/2100777806045327372)
[^3]: [𝕏 post by @S1r1u5_](https://x.com/S1r1u5_/status/2100777825951433200)
[^4]: [𝕏 post by @AnthropicAI](https://x.com/AnthropicAI/status/2100684274114699295)
[^5]: [𝕏 post by @kimmonismus](https://x.com/kimmonismus/status/2100703850630205949)
[^6]: [𝕏 post by @ValsAI](https://x.com/ValsAI/status/2100676214403088478)
[^7]: [𝕏 post by @ValsAI](https://x.com/ValsAI/status/2100676216571547652)
[^8]: [𝕏 post by @ValsAI](https://x.com/ValsAI/status/2100676219272597686)
[^9]: [𝕏 post by @ValsAI](https://x.com/ValsAI/status/2100676221487219185)
[^10]: [𝕏 post by @mirrokni](https://x.com/mirrokni/status/2100660772393320476)
[^11]: [𝕏 article by @GoogleAIStudio](https://x.com/i/article/2099964776117899264)
[^12]: [𝕏 post by @OpenAI](https://x.com/OpenAI/status/2100679994305630562)
[^13]: [𝕏 post by @OpenAI](https://x.com/OpenAI/status/2100679996633452747)
[^14]: [𝕏 post by @OpenAI](https://x.com/OpenAI/status/2100680000072773702)
[^15]: [𝕏 post by @ValsAI](https://x.com/ValsAI/status/2100714091845497092)
[^16]: [𝕏 post by @ruima](https://x.com/ruima/status/2100487077318336955)
[^17]: [𝕏 post by @arcee_ai](https://x.com/arcee_ai/status/2100230847907459094)
[^18]: [𝕏 post by @Alibaba_Qwen](https://x.com/Alibaba_Qwen/status/2100785962414702599)
[^19]: [𝕏 post by @EpochAIResearch](https://x.com/EpochAIResearch/status/2100704765332394255)
[^20]: [𝕏 post by @nrehiew_](https://x.com/nrehiew_/status/2100716134261744089)
[^21]: [𝕏 post by @dair_ai](https://x.com/dair_ai/status/2100695797847466435)