Skip to content
ReframedNewsLeft · Right · Centered
Quick ReadArchiveAboutSign in
Briefing stories›Technology & Society

Anthropic limits access to Claude Mythos Preview through cybersecurity consortium

Story date · Tuesday, April 7, 2026Technology & SocietyWell-covered

The facts

  • Anthropic announced a cybersecurity initiative called Project Glasswing centered on its unreleased AI model Claude Mythos Preview.
  • Anthropic is limiting access to Claude Mythos Preview instead of making it generally available to the public.
  • Project Glasswing includes more than 40 participating organizations or companies.
  • Named partners in the initiative include major technology companies such as Apple, Amazon, Microsoft, Google, Nvidia, Cisco, CrowdStrike and Palo Alto Networks.
  • Anthropic says Claude Mythos Preview is designed to be used for defensive cybersecurity work, including finding and helping patch software vulnerabilities.
  • Anthropic says the model has identified thousands of high-severity or serious software vulnerabilities.
  • Reports on the announcement say Claude Mythos Preview found a 27-year-old vulnerability in OpenBSD.

Context

What is Project Glasswing?

Project Glasswing is Anthropic's new cybersecurity initiative, which gives a limited set of partner organizations access to Claude Mythos Preview for defensive security work on critical software and infrastructure NYT,Reuters,Bloomberg Business.

Why isn't Anthropic releasing Claude Mythos Preview publicly?

Anthropic says it is restricting the model because of concerns that its ability to find and exploit software weaknesses could be misused by bad actors, so it is limiting use to trusted partners for defensive purposes NYT,CNBC,Business Insider.

What has Anthropic said the model can do?

Anthropic says Claude Mythos Preview is especially strong at identifying serious software vulnerabilities and has already uncovered thousands of severe flaws, including long-undetected bugs such as a 27-year-old issue in OpenBSD T-online.de,Investing.com,newsORF.at.

See this differently than someone you know would?

The receipts — 100 sources

Wire services (3)

APCointelegraphAnthropic Holds Back Claude Mythos After it Found Thousands ...
APMediapartAnthropic reporte la sortie de sa nouvelle IA, trop dangereu...
ReutersReutersAnthropic touts AI cybersecurity project with Big Tech partn...

Other coverage (97)

Corriere della SeraClaude Mythos, il modello di Anthropic che spaventa: "Per or...
Notícias ao MinutoAnthropic lança modelo de IA criado para melhorar cibersegur...
newsORF.atNeue Anthropic-KI wohl zu gefährlich für Veröffentlichung
YourStory.comAnthrophic has a potent AI model on hand, but it isn't ready...
NotebookcheckZu riskant: Darum macht Antrophic seine neue Cyber-KI nicht ...
Euronews EnglishWhat happens when AI becomes too powerful? Anthropic is find...
Le journal du netClaude Mythos : le jour où l'IA a changé de catégorie
Аргументы и фактыAnthropic ограничила доступ к новой нейросети из-за ее опасн...
ComputerBaseAutonom Exploits entwickeln: Anthropics neues Modell ist so ...
manager magazinAnthropic: Neue KI "Mythos" zu gefährlich für die Öffentlich...
Business StandardAnthropic unveils Project Glasswing to boost cybersecurity w...
AxiosAnthropic's newest AI model could wreak havoc. Most in power...
早报Anthropic与40多家科企合作 测试AI模型网络攻击能力
TechRadar'A new frontier model trained by Anthropic that we believe c...
T-online.deOpenAI-Konkurrent: Wie diese Software uralte Schwachstellen ...
WinFuture.deAnthropic: Neues KI-Modell Mythos zu gefährlich für die Öffe...
Merkur.deDiese KI darf man nicht nutzen: Claude Mythos bleibt unter V...
Economic TimesClaude Mythos, Anthropic AI capable of hacking any software,...
La RazónAnthropic tiene una versión tan potente de Claude que no pue...
The Times of IndiaAnthropic is not releasing its newest AI model, Mythos, to p...
Business InsiderAnthropic hält neues KI-Modell zurück - das ist der Grund
europa pressAnthropic impulsa el Proyecto Glasswing para reforzar la...
Blick.chMythos: Anthropic baut KI - und hält sie unter Verschluss
N-tvBesorgniserregende Fähigkeiten: Neues KI-Modell entdeckt sei...
FanpageAnthropic svela Project Glasswing: l'obiettivo è trovare con...
DigitClaude Mythos Preview: Everything to know about world's most...
Clubic.comAvec Claude Mythos preview, Anthropic détecte des failles ze...
ХабрВ тестах Claude Mythos сбежал из "тюрьмы", а затем рассказал...
The New Indian ExpressLatest Anthropic AI model finds cracks in software defences
01netEffrayé par un " effet secondaire " du nouveau Claude, Anthr...
Kronen ZeitungAnthropic-KI fand "Tausende" Sicherheitslücken
The Indian ExpressWhat is Claude Mythos, and why is Anthropic limiting its rol...
XatakaClaude Mythos es un modelo de IA tan potente que da miedo. A...
The Indian ExpressAnthropic launches Project Glasswing to test advanced AI for...
FrandroidClaude Mythos, une IA vraiment trop puissante pour notre pro...
Spiegel OnlineKI findet tief versteckte Software-Schwachstellen
SudOuest.frCybersécurité : Anthropic reporte la sortie de sa nouvelle I...
The Financial ExpressAnthropic not releasing Claude Mythos and reason is alarming...
NDTVAnthropic's Claude Mythos AI Exposes Critical Software Secur...
heise onlineAnthropics neues KI-Modell Mythos: Zu gefährlich für die Öff...
Cadena 3 ArgentinaAnthropic presenta su potente modelo de IA Mythos para ciber...
FoneArenaAnthropic unveils Project Glasswing to strengthen AI-driven ...
NZ HeraldSecurity reckoning: New Claude model kept from public, used ...
TimesNowAnthropic Launches 'Project Glasswing', AI That Can Find Sof...
ETCISO.inProject Glasswing: Tech giants unite to secure critical soft...
India TodayMeet Claude Mythos, Anthropic AI so powerful it can hack any...
DigitAnthropic launches Project Glasswing to fight AI-driven cybe...
Asian News International (ANI)Anthropic announces 'Project Glasswing' in alliance with tec...
工商時報合作而非取代 Anthropic攜手資安巨頭強化安全
GizmodoAnthropic's New Model Is So Scarily Powerful It Won't Be Rel...
FirstpostProject Glasswing: Anthropic teams up with Apple, Google to ...
Vorarlberg OnlineKI-Modell von Anthropic entdeckt alte Softwareschwachstellen
The Globe and MailAnthropic claims its new AI model, Mythos, is a cybersecurit...
Free Malaysia TodayLatest Anthropic AI model finds cracks in software defences
Freie PresseKI findet seit Jahren schlummernde Software-Schwachstellen
DIE WELTKI findet seit Jahren schlummernde Software-Schwachstellen -...
Der TagesspiegelKünstliche Intelligenz: KI findet seit Jahren schlummernde S...
HandelsblattKünstliche Intelligenz: KI findet seit Jahren schlummernde S...
News aus OWLKI findet seit Jahren schlummernde Software-Schwachstellen
stuttgarter-nachrichten.deKünstliche Intelligenz: KI findet seit Jahren schlummernde S...
The New York TimesOpinion | Anthropic's Restraint Is a Terrifying Warning Sign
MoneyDJ理財網合作而非取代 Anthropic攜手資安巨頭強化安全-MoneyDJ理財網
Australian Financial ReviewAnthropic pits bot against bot in AI cyberwar with powerful ...
ForbesFive Reasons Anthropic Kept Its Cybersecurity Breakthrough I...
TheRegister.comAnthropic Mythos model can find and exploit 0-days
MorningstarCrowdStrike, Palo Alto Networks shares pop as cybersecurity ...
Olhar Digital - O futuro passa primeiro aquiAnthropic lança novo modelo de IA para fortalecer cibersegur...
EL MUNDOLa IA más avanzada de Anthropic es tan peligrosa que, de mom...
GizmodoAnthropic Launches 'Project Glasswing' to Stealthily Spot Cy...
Business InsiderAnthropic says its latest AI model is too powerful for publi...
Numerama.comClaude Mythos : le nouveau modèle IA d'Anthropic jugé trop d...
Le Figaro.frAnthropic identifie "des milliers de failles informatiques" ...
Il Sole 24 OREAnthropic dà vita a Project Glasswing, l'intelligenza artifi...
stuttgarter-nachrichten.deKünstliche Intelligenz: KI findet seit Jahren schlummernde S...
RP OnlineKünstliche Intelligenz: KI findet seit Jahren schlummernde S...
Economic TimesAnthropic touts AI cybersecurity project with Big Tech partn...
Pulse24.com Latest Anthropic AI model finds cracks in software defences
Fast CompanyAnthropic will use its biggest, baddest AI model to protect ...
WiredAnthropic Teams Up With Its Rivals to Keep AI From Hacking E...
9to5MacAnthropic unveils powerful Mythos AI model, working with App...
ХабрНовейший Claude Mythos нашел дыру возрастом 27 лет в OpenBSD
FortuneAnthropic is giving some firms access to Claude Mythos to bo...
La Presse.caAnthropic | Des milliers de failles de cybersécurité repérée...
Investing.com РоссияAnthropic запускает Project Glasswing для защиты ПО От Inves...
Investing.com ItaliaAnthropic lancia Project Glasswing per proteggere software c...
Investing.com FranceAnthropic lance le Projet Glasswing pour sécuriser les logic...
Investing.comAnthropic launches Project Glasswing to secure critical soft...
TVA NouvellesDes milliers de failles de cybersécurité repérées par la nou...
Inc.Anthropic's Claude Mythos Is So Powerful, It 'Could Reshape ...
CNAAnthropic touts AI cybersecurity project with Big Tech partn...
ZDNetApple, Google, and Microsoft join Anthropic's Project Glassw...
The New York TimesAnthropic Claims Its New A.I. Model, Mythos, Is a Cybersecur...
AxiosAnthropic withholds Mythos Preview model because it's hackin...
CNBCAnthropic limits Mythos AI rollout over fears hackers could ...
The Wall Street JournalAnthropic Set to Preview Powerful 'Mythos' Model to Ward Off...
Bloomberg BusinessAnthropic Lets Apple, Amazon Test More Powerful Mythos AI Mo...
The VergeAnthropic is launching a new AI model for cybersecurity

A news briefing built to show both sides.

The day’s biggest stories in one short brief, built to show both sides and the competing values behind the headlines. Free in your inbox.

← Earlier in this briefing
Australia charges former SAS soldier with five war crime murder counts over Afgh...
Australian authorities have charged a former Special Air Service soldier with five counts of war crime murder over the...
Rights & Justice
Next in this briefing →
Brazil adds BYD to labor ministry registry over slavery-like working conditions ...
Brazil's Labor Ministry has added BYD to its registry of employers linked to conditions analogous to slavery, following...
Rights & Justice

Continue Reading

More in Technology & Society

Greece says it will ban social media access for children under 15 from January 2027

Greek Prime Minister Kyriakos Mitsotakis said Greece will ban access to social media for children under 15 starting on...

Technology & Society
Related this week

Canada pledges up to C$560 million to Brazilian-led tropical forest fund

Canada pledged up to C$560 million (US$392.5 million) over four years to the Brazilian-led Tropical Forests Forever...

Science & Climate
From this briefing

Judge says CBP improperly canceled Harvard researcher’s visa in frog embryo smuggling case

A federal judge ruled Tuesday that Customs and Border Protection improperly canceled the visa of Kseniia Petrova, a...

Rights & Justice
Back to all stories

A news briefing built to show both sides.

The day’s biggest stories in one short brief, built to show both sides and the competing values behind the headlines. Free in your inbox.

ReframedNews

A daily news briefing on the day’s top stories, built to show both sides.

Written by AI from news reporting. We review published briefings. Who writes this →

Navigate

Quick ReadArchiveSettings

Company

AboutHow it worksSkylark CreationsSign InTerms of ServicePrivacy Policy

© 2026 Reframed.News

Made by Skylark Creations