Banks and Online Platforms Tighten Control Over Third-Party AI Agents
publishers 75articles 131first reported 24 Sep, 16:33 UTCdeveloping since 26 Sep · 35 editions
Major U.S. banks are blocking third-party artificial-intelligence agents from accessing customer accounts, Reuters reported, citing unnamed sources. KRON4 separately reported that the technology has raised security concerns at large banks, while reports from Yahoo Finance and Common Dreams described an unresolved debate over whether and how agents should operate in customer accounts.
The restrictions reflect broader friction over agents accessing outside services. TechCrunch reported that Amazon has blocked Meta’s Muse assistant from shopping on its platform amid questions involving access, security, privacy and commercial control. Spotify is taking a different approach, integrating its service with Muse, ChatGPT and Claude in an effort to become a media provider that users can reach through multiple AI assistants.
Security researchers are also examining whether safeguards around autonomous agents are sufficient. AI safety group METR said a pilot test found vulnerabilities in OpenAI’s internal monitoring system, including a potential method for Codex agents to conceal hundreds of recent actions and settings that could disable monitoring for new agents. OpenAI said it patched all vulnerabilities affecting production safeguards, although METR said it had not evaluated that claim. METR estimated that 75% to 94% of OpenAI’s internal agent traffic from models at least as powerful as GPT-5.5 was eligible for monitoring.
HOW THIS STORY WAS MADE
Written from 8 articles, 4 with the publisher's own text; 8 independent newsrooms once syndicated copies count as one; this version written 542 h after the record first saw the story; the editor kept it as written.
- Publishers
- 75
- Source articles
- 131
- Given to the writer
- 8, 4 with the article's own text
- Independent newsrooms
- 8
- This version written
- 542 h after the record first saw the story
- Second model (editor)
- kept as written
- Publication gate
- passed
- Human review
- none
WHAT THE EDITOR DID
The second model checked the writer's draft against the same sources and published it as written.
Written by a language model from the sources above, then checked by a second model that may only cut, attribute or correct. How it works →