# OpenAI quietly withdraws from pushing for chip export restrictions to China, report says

Digest AI · Policy & Regulation · published 2026-10-20T22:24:00Z

Canonical: https://digestai.news/story/openai-quietly-withdraws-from-pushing-for-chip-export-restrictions-to

## Summary

A hacker using AI agents breached more than 100 corporate websites across the US, Canada, Australia and Japan, costing roughly $5,000 total – about $50 per intrusion – according to cybersecurity researcher Snorre Fagerland of NorForge and a Reuters report. The campaign leveraged the open‑source Cairn platform and targeted firms such as Deere, SharkNinja and Cummins, though the extent of data loss is still unknown.

Microsoft announced new "Project Perception" controls aimed at treating autonomous AI agents as insider threats. The features, including dynamic permission granting and continuous risk scoring, are now available to Microsoft 365 E5 customers with a broader rollout to mid‑tier licenses planned for the first quarter of 2027, as detailed in the company’s security blog.

Politico reported that OpenAI has stopped advocating strict chip export restrictions on China. The shift follows Nvidia’s $3 billion equity investment in OpenAI and a $105 billion data‑center guarantee in Ohio, suggesting the company’s policy is influenced by its deepening relationship with the chip maker. Meanwhile, Meta completed phase 1 of its Prometheus supercluster and announced GPT‑6.1 Sol achieving 97.9% accuracy on COBOL tasks, highlighting rapid advances in both infrastructure and model capabilities.

## Key points

- AI agents breached over 100 websites for about $5,000 total, roughly $50 per intrusion, using the open‑source Cairn platform (Reuters).
- Microsoft announced Project Perception controls for malicious AI agents, now in Microsoft 365 E5 and slated for mid‑tier licenses in Q1 2027.
- OpenAI reportedly stopped advocating strict chip export limits to China after Nvidia’s $3 billion equity investment and a $105 billion Ohio data‑center deal.

## Why it matters

Low‑cost AI‑driven attacks lower the barrier for cybercrime, prompting firms like Microsoft to embed agent‑control tools, while OpenAI’s policy shift tied to Nvidia financing reshapes US‑China AI competition and could affect future export regulations.

## Sources

1. [AI Agents Breach Over 100 Companies, OpenAI Quietly Withdraws from China Export Restrictions | AI News 10/20](https://note.com/chic_hound2052/n/n775ef0d4b94e?hl=en) (note.com, 2026-10-20)

Part of the developing story: [AI Security and Export Policy Shifts](https://digestai.news/thread/microsoft-outlines-project-perception-to-guard-agentic-insiders) (2 stories)

## Cite

Digest AI, "OpenAI quietly withdraws from pushing for chip export restrictions to China, report says", 20 October 2026, https://digestai.news/story/openai-quietly-withdraws-from-pushing-for-chip-export-restrictions-to

---

Written by Digest AI's editorial model from the linked sources; the sources are the record. Headlines, digests and key points are written by Digest AI and may be quoted with a link to the story page. Linked articles belong to their publishers. Terms: https://digestai.news/terms#reuse
JSON: https://digestai.news/story/openai-quietly-withdraws-from-pushing-for-chip-export-restrictions-to.json