Wire report

‘Not perfectly aligned’ with human values: Anthropic admits security failures behind AI hacking incidents

US owner of Claude chatbot previously said its models had hacked three organisations during testing The US startup behind the Claude chatbot has admitted a series of hacking incidents involving its models reflected a “failure of operational security” and said it has tightened its testing procedures. Anthropic revealed in July that its models had accessed the open internet three times and gained unauthorised access to the systems of three organisations. Continue reading...

Source-feed image associated with ‘Not perfectly aligned’ with human values: Anthropic admits security failures behind AI hacking incidents
Source-feed image associated with the linked report: ‘Not perfectly aligned’ with human values: Anthropic admits security failures behind AI hacking incidents.Credit: The Guardian Source-feed thumbnail displayed with attribution and outbound source link; VINI does not claim ownership or republish the third-party article body. Image source Cached source-feed image shown for continuity with attribution and an outbound source link; VINI does not claim third-party image authorship or republish the third-party article body.
Reading time1 min

coverage / Wire report

Reader toolsFollow the reporting.

Get updates, read source context, send useful records, share the story, or support the reporting work from the reading page.

FollowGet story updatesBriefs and topic returnsContextOpen background1 public sourceContributeSend recordsDocuments, dates, photosSupportFund reportingReader-backed workShareCopy story URLvininews.com
Why it mattersTechnology

US owner of Claude chatbot previously said its models had hacked three organisations during testing The US startup behind the Claude chatbot has admitted a series of hacking incidents involving its models reflected a “failure of operational security” and said it has tightened its testing procedures. Anthropic revealed in July that its models had accessed the open internet three times and gained unauthorised access to the systems of three organisations. Continue reading...

What to know1 source

Check the original link, updates, and responses when a detail is contested.

Keep readingtech

Open topic or search related wording such as records, sources, agencies, dates, and locations.

What happened

According to The Guardian’s source item, ‘Not perfectly aligned’ with human values: Anthropic admits security failures behind AI hacking incidents, US owner of Claude chatbot previously said its models had hacked three organisations during testing The US startup behind the Claude chatbot has admitted a series of hacking incidents involving its models reflected a “failure of operational security” and said it has tightened its testing procedures. Anthropic revealed in July that its models had accessed the open internet three times and gained unauthorised access to the systems of three organisations. Continue reading…

Context

The development sits in VINI’s Technology file for readers following technology, science, product policy, markets, infrastructure, and the public consequences of innovation. The original report is linked so readers can check the source account, follow later updates, and compare new coverage against the first published record. The source item is dated 2026-09-01T15:18:10+00:00.

What to watch

Open questions include whether primary sources issue follow-up statements, whether local or market impacts become clearer, and whether additional reporting changes the timeline or adds material context.

Source

Primary source: ‘Not perfectly aligned’ with human values: Anthropic admits security failures behind AI hacking incidents via The Guardian. VINI cites and links the source; it does not reproduce the publisher’s full article text without rights clearance.

Keep following

This file can keep developing

vininews.com uses reader tips, public records, right-of-reply requests, corrections, and follow-up reporting to keep important stories current.

SubscribeGet the next updateSend recordsShare documents or leadsRespondRequest comment or replyDonateSupport reporting costs

Support and subscriptions never buy coverage, placement, suppression, or corrections.

This VINI report keeps the original publisher link available and does not republish third-party article bodies without rights clearance. 1 reference listed.

Source links

Reader comments

Moderated discussion

Account access

Comments are open to authenticated approved accounts, screened for spam and abuse, and published only after newsroom moderation unless editors change the story control.

Loading comments.

No approved comments yet.

Substantive, civil comments can be submitted by approved account holders.