Apple flags autonomous AI risks as industry debate widens
publishers 131articles 443first reported 23 Sep, 14:00 UTCdeveloping since 30 Sep · 45 editions
Apple warned in its annual financial filing Tuesday that autonomous artificial intelligence agents could expose the company to lawsuits and reputational damage by behaving in unintended or unpredictable ways.
The company said agents can interact with systems they may not be authorized to access, alter code, send communications and complete transactions. It added that such systems could take actions that were not directed or authorized, while Apple’s ability to oversee third-party agents may be limited. Apple has also announced changes to full-disk access on Macs to reduce risks associated with AI agents.
Chief executive John Ternus said on an earnings call that agentic tools were promising but that users should understand what access they grant them. He said Apple saw room to expand Siri’s capabilities while protecting security and privacy.
The warning comes amid a broader dispute over how quickly increasingly capable AI should be developed and released. Geoffrey Hinton, a Nobel prize-winning computer scientist, told the Guardian that AI could cause a major incident involving financial or power systems this year or next, while also acknowledging its potential benefits.
Meanwhile, DeepSeek engineer Gao Xi defended open-source development at the World Internet Conference in Wuzhen, arguing that shared innovation could accelerate artificial general intelligence despite heightened safety risks. Gao said he hoped AGI would arrive as quickly as possible.
HOW THIS STORY WAS MADE
Written from 8 articles, 7 with the publisher's own text; 7 independent newsrooms once syndicated copies count as one; this version written 834 h after the record first saw the story; the editor revised it.
- Publishers
- 131
- Source articles
- 443
- Given to the writer
- 8, 7 with the article's own text
- Independent newsrooms
- 7
- This version written
- 834 h after the record first saw the story
- Second model (editor)
- revised
- Publication gate
- passed
- Human review
- none
WHAT THE EDITOR DID
The second model checked the writer's draft against the same sources and changed it before publication. Struck words were in the draft; underlined words are its corrections.
WHAT IT FLAGGED editor revise: Correct the status of Apple’s full-disk-access changes to match the source. - "Apple has also changed full-disk access on Macs to reduce risks associated with AI agents.": The source says Apple announced changes to full-disk access, not that the changes had already been made.
THE DRAFT AND THE CHANGES
Apple flags autonomous AI risks as industry debate widens
Apple warned in its annual financial filing Tuesday that autonomous artificial intelligence agents could expose the company to lawsuits and reputational damage by behaving in unintended or unpredictable ways.
The company said agents can interact with systems they may not be authorized to access, alter code, send communications and complete transactions. It added that such systems could take actions that were not directed or authorized, while Apple’s ability to oversee third-party agents may be limited. Apple has also changedannounced changes to full-disk access on Macs to reduce risks associated with AI agents.
Chief executive John Ternus said on an earnings call that agentic tools were promising but that users should understand what access they grant them. He said Apple saw room to expand Siri’s capabilities while protecting security and privacy.
The warning comes amid a broader dispute over how quickly increasingly capable AI should be developed and released. Geoffrey Hinton, a Nobel prize-winning computer scientist, told the Guardian that AI could cause a major incident involving financial or power systems this year or next, while also acknowledging its potential benefits.
Meanwhile, DeepSeek engineer Gao Xi defended open-source development at the World Internet Conference in Wuzhen, arguing that shared innovation could accelerate artificial general intelligence despite heightened safety risks. Gao said he hoped AGI would arrive as quickly as possible.
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 →