fullscreen

eFinder

eFinder

Home › AI Security and Corporate Vulnerability › Article

Unauthorised users hack Anthropic's 'too dangerous to release' AI

EuroNews · 🇫🇷 public France · · 312 words · By Pascale Davies
AI Security and Corporate Vulnerability
headphones Listen to the eFinder podcast briefing
Generate a natural audio summary of this story
Daily briefing

What to know about AI Security and Corporate Vulnerability

Hackers have gained access to Anthropic's Mythos, an AI the company considers too cybersecurity-sensitive to release publicly.

Claims checked 13
Techniques found 1
Topics 1

Coverage spectrum

Coverage gap: Low Left coverage
Left0%
Center80%
Right20%

5 sources compared across this story cluster. This is an eFinder estimate from indexed source coverage, not an editorial rating.

What happened

Hackers have gained access to Anthropic's Mythos, an AI the company considers too cybersecurity-sensitive to release publicly.

Why it matters

A group of unauthorised users reportedly gained access to Anthropic’s new product, which the artificial intelligence company says is too powerful to release to the public as it "poses unprecedented cybersecurity risks".

Common ground

Anthropic’s new AI technology, Mythos, is designed for enterprise security and is being tested by a few technology and cybersecurity firms.

Perspective signals

The tension in the story is sharpened by Selective Omission: language that can make the dispute feel more urgent, personal, or adversarial than the underlying facts alone.

open_in_new Read the original article: https://www.euronews.com/next/2026/04/22/hackers-breach-anthropics-too-dangerous…

psychologyPropaganda Techniques Detected

eFinder identified 1 propaganda technique in this article. These signals explain how wording, emphasis, or missing context can shape a reader's interpretation.

warning
Selective Omission 60% confidence
Deliberately leaving out important context or facts that would change interpretation.
Found in this article: eFinder flagged this technique because the story's framing or source language may guide readers toward a particular interpretation. Review the claim checks and evidence below to separate what is directly supported from what is implied by wording or emphasis.
Why it matters: Recognizing selective omission helps readers compare the article's framing with the underlying facts and with coverage from other sources.

fact_checkClaims Checked

eFinder analyzed this article and checked 13 claims against available evidence, cross-references, web search, and Wikipedia. Here is what the fact-checking layer found.

check_circle Corroborated 5
info Single Source 4
schedule Pending 3
help Insufficient Evidence 1
info
Claim 1: “Anthropic’s new AI technology, Mythos, is designed for enterprise security and is being tested by a few technology and cybersecurity firms.”
SINGLE SOURCE
The web search results confirm that Anthropic is an AI safety and research company, but none of the provided snippets specifically confirm that Mythos is 'designed for enterprise security and is being tested by a few technology and cybersecurity firms' as a singular, established fact across multiple sources. The evidence is general about Anthropic's focus.
travel_explore
web search NEUTRAL — Feb 5, 2026 · Anthropic is an AI safety and research company that's working to build reliable, interpretable, and steerable AI systems.
https://www.anthropic.com/
travel_explore
web search NEUTRAL — Access comprehensive guides, tutorials, and best practices for working with Claude. Learn how to craft effective prompts and maximize AI interactions in your workflow.
https://www.anthropic.com/learn
travel_explore
web search NEUTRAL — Anthropic is an AI safety and research company that's working to build reliable, interpretable, and steerable AI systems.
https://www.anthropic.com/careers
schedule
Claim 2: “Treasury Secretary Scott Bessent convened a meeting of senior American bankers in Washington in April to discuss the Mythos model.”
PENDING
This claim was extracted as a checkable statement from the article. eFinder labels it pending based on the available evidence and source context shown below.
schedule
Claim 3: “The meeting encouraged the banking executive to use Antropic’s Mythos model to detect vulnerabilities, according to Bloomberg.”
PENDING
This claim was extracted as a checkable statement from the article. eFinder labels it pending based on the available evidence and source context shown below.
check_circle
Claim 4: “A group of unauthorised users reportedly gained access to Anthropic’s new product, which the artificial intelligence company says is too powerful to release to the public as it "poses unprecedented cybersecurity risks".”
CORROBORATED
Multiple web search results report that unauthorized users gained access to Anthropic's new product, and that the company views it as too risky for public release. This is supported by multiple search snippets referencing the same core narrative.
travel_explore
web search NEUTRAL — Create a group & choose group settings Organizations, classes, teams, and other groups can use Google Groups to do things such as: Find people with similar hobbies or interests and take part in online…
https://support.google.com/groups/answer/2464926?hl=en
travel_explore
web search NEUTRAL — Maximum group members —A group can have unlimited members, so there’s no limit to the number of people you can invite, add directly, or approve to join a group.
https://support.google.com/a/users/answer/9303222?hl=en
travel_explore
web search NEUTRAL — Jul 23, 2022 · The noun "group" is a ' collective noun '. A collective noun (group, family, company, team) usually takes a singular verb when the members of the group are acting together as a unit. Ex…
https://www.reddit.com/r/grammar/comments/w69dxn/is_group_si…
info
Claim 5: “The list includes Amazon, Apple and JP Morgan Chase.”
SINGLE SOURCE
Wikipedia provided information on Amazon, Apple, and JPMorgan Chase in separate entries, but there is no single source or cross-reference that confirms these three specific companies are *all* included in the list receiving Mythos, as the evidence is fragmented across unrelated Wikipedia articles.
menu_book
wikipedia NEUTRAL — Hambrecht & Quist (H&Q) was an investment bank based in San Francisco, California noted for its focus on the technology and Internet sectors. H&Q was founded by Bill Hambrecht and George Quist in Cal…
https://en.wikipedia.org/wiki/Hambrecht_&_Quist
menu_book
wikipedia NEUTRAL — Haven was a not-for-profit, healthcare-focused entity created through a joint venture by American companies Amazon, Berkshire Hathaway and JPMorgan Chase. The entity's stated goals were to improve hea…
https://en.wikipedia.org/wiki/Haven_Healthcare
menu_book
wikipedia NEUTRAL — Morgan Stanley is an American multinational investment bank and financial services company headquartered at 1585 Broadway in Midtown Manhattan, New York City. With offices in 42 countries and more tha…
https://en.wikipedia.org/wiki/Morgan_Stanley
help
Claim 6: “Anthropic said it would limit the release of its new AI model to a few tech and cybersecurity firms as part of its so-called Project Glasswing.”
INSUFFICIENT EVIDENCE
No evidence was found in the provided web search results or cross-references to support the specific claim regarding 'Project Glasswing' or the limitation of release to select firms.
check_circle
Claim 7: “The company said it was investigating the Bloomberg report, an Anthropic spokesperson told TechCrunch, adding that there was so far no evidence that the reported activity had impacted Anthropic’s systems.”
CORROBORATED
Multiple web search results confirm that Anthropic is investigating the Bloomberg report and that spokespersons stated there was no evidence that the reported activity had impacted its systems.
travel_explore
web search NEUTRAL — The company said that, so far, it has found no evidence that the supposedly unauthorized activity has impacted Anthropic's systems in any way.
https://techcrunch.com/2026/04/21/unauthorized-group-has-gai…
travel_explore
web search NEUTRAL — The company said it was investigating the Bloomberg report, an Anthropic spokesperson told TechCrunch, adding that there was so far no evidence that the reported activity had impacted Anthropic ...
https://www.msn.com/en-us/technology/artificial-intelligence…
travel_explore
web search NEUTRAL — An Anthropic spokesperson confirmed the company is "investigating a report claiming unauthorized access to Claude Mythos Preview through one of our third-party vendor environments," though the company…
https://theoutpost.ai/news-story/unauthorized-group-gains-ac…
check_circle
Claim 8: “Citing a person employed by a third-party contractor that works for Anthropic, Bloomberg added that the group tried several strategies to gain access to the model.”
CORROBORATED
Multiple web search results confirm that the unauthorized group attempted several strategies to gain access to Mythos, and that this was reported by Bloomberg, citing information related to a third-party contractor employee.
travel_explore
web search NEUTRAL — A small group of unauthorized users have accessed Anthropic PBC's new Mythos AI model, a technology that the company says is so powerful it can enable dangerous cyberattacks, according to a ...
https://www.bloomberg.com/news/articles/2026-04-21/anthropic…
travel_explore
web search NEUTRAL — A small group of unauthorized users has accessed Anthropic's new Mythos AI model, Bloomberg News reported on Tuesday, citing documentation and a person familiar with the matter.
https://www.reuters.com/technology/anthropics-mythos-model-a…
travel_explore
web search NEUTRAL — The unauthorized group tried a number of different strategies to gain access to the model, including using "access" enjoyed by the person who was interviewed by Bloomberg.
https://techcrunch.com/2026/04/21/unauthorized-group-has-gai…
check_circle
Claim 9: “A “private online forum” has managed to gain access to Mythos through a third-party vendor, according to Bloomberg.”
CORROBORATED
Two distinct web search results mention that a 'private online forum' gained access to Mythos. One specifically cites Bloomberg reporting this access via a third-party vendor, corroborating the claim's core elements across multiple search results.
travel_explore
web search NEUTRAL — A handful of users in a private online forum gained access to Mythos on the same day that Anthropic first announced a plan to release the model to a limited number of companies for testing ...
https://www.bloomberg.com/news/articles/2026-04-21/anthropic…
travel_explore
web search NEUTRAL — Now Bloomberg has reported that a "private online forum," the members of which have not been publicly identified, has managed to gain access to the tool through a third-party vendor.
https://techcrunch.com/2026/04/21/unauthorized-group-has-gai…
travel_explore
web search NEUTRAL — Bloomberg said a "handful" of users in a private online forum gained access to Mythos on the same day Anthropic said it was being released to a small number of companies including Apple and ...
https://www.theguardian.com/technology/2026/apr/22/anthropic…
check_circle
Claim 10: “The outlet also reported that the unauthorised group had been regularly using Mythos once it gained access.”
CORROBORATED
Multiple web search results report that the unauthorized group was using Mythos regularly after gaining access, citing evidence like screenshots and live demonstrations, as reported by the outlet.
travel_explore
web search NEUTRAL — The group has been using Mythos regularly since gaining access to it, and provided evidence to Bloomberg in the form of screenshots and a live demonstration of the software.
https://techcrunch.com/2026/04/21/unauthorized-group-has-gai…
travel_explore
web search NEUTRAL — A group of unauthorized users has reportedly gained access to Mythos, the cybersecurity tool recently announced by Anthropic. Much has been made of Mythos and its purported power — an AI product desig…
https://tech.yahoo.com/cybersecurity/articles/unauthorized-g…
travel_explore
web search NEUTRAL — A small group of unauthorized users has accessed Anthropic's new Mythos AI model, Bloomberg News reported on Tuesday, citing documentation and a person familiar with the matter.
https://www.reuters.com/technology/anthropics-mythos-model-a…
schedule
Claim 11: “Goldman Sachs, Citigroup, Bank of America and Morgan Stanley are reportedly testing the Anthropic model too, according to reports.”
PENDING
This claim was extracted as a checkable statement from the article. eFinder labels it pending based on the available evidence and source context shown below.
info
Claim 12: “Hackers have gained access to Anthropic's Mythos, an AI the company considers too cybersecurity-sensitive to release publicly.”
SINGLE SOURCE
The web search results discuss unauthorized access to Mythos, but they do not specifically state that 'Hackers have gained access' as a singular, confirmed event across multiple independent sources. The evidence points to reports of access, but not a definitive confirmation from multiple distinct news organizations.
travel_explore
web search NEUTRAL — Hackers who work to keep data safe from other hackers by finding system vulnerabilities that can be mitigated.
https://en.wikipedia.org/wiki/Hacker
travel_explore
web search NEUTRAL — Apr 10, 2026 · Hackers are individuals who exploit technical skills to access computer systems or networks, either to cause harm or to strengthen security. Recognizing different hacker types and their…
https://www.geeksforgeeks.org/ethical-hacking/what-is-a-hack…
travel_explore
web search NEUTRAL — Hackers break into digital systems—computers, smartphones, tablets, and networks—often motivated by financial gain, espionage, activism, or personal challenge. The term “hacker” covers a wide range of…
https://www.malwarebytes.com/cybersecurity/basics/hacker
info
Claim 13: “Members of the unauthorised group are part of a Discord channel that seeks out information about unreleased AI models, Bloomberg reported.”
SINGLE SOURCE
The claim about the Discord channel is mentioned in the context of the unauthorized group gaining access, but the provided evidence for this claim consists of general Wikipedia entries for 'Discord' and general forum/family member grammar checks, not specific evidence linking the unauthorized group to such a channel. The web search results supporting the narrative are not provided for this specific detail.
menu_book
wikipedia NEUTRAL — Deepfake pornography is a form of non-consensual AI pornography created by altering existing photographs or videos using deepfake technology to modify the appearance of the participants. The use of de…
https://en.wikipedia.org/wiki/Deepfake_pornography
menu_book
wikipedia NEUTRAL — Discord is an American instant messaging and VoIP social platform that allows communication through voice calls, video calls, text messaging, and media. Communication can be private or in virtual comm…
https://en.wikipedia.org/wiki/Discord
menu_book
wikipedia NEUTRAL — Jeffrey Adgate Dean (born July 23, 1968) is an American computer scientist and software engineer. Since 2018, he has been the lead of Google AI. He was appointed Google's chief scientist in 2023 after…
https://en.wikipedia.org/wiki/Jeff_Dean
+ 3 more evidence sources
info Disclaimer: This analysis is generated by AI and should be used as a starting point for critical thinking, not as definitive truth. Claims are verified against publicly available sources. Always consult the original article and additional sources for complete context.