Observed Signal · Jul 29, 2026 · Research Finding · Source: t3n · Impact: 4/5 · Sentiment: Neutral
Anthropic AI Finds Flaws in Weakened AES Variant
Anthropic's Claude Mythos Preview reportedly discovered vulnerabilities in a weakened variant of the AES encryption standard and produced an improved attack on Hawk, a post-quantum candidate. Reporting says Mythos found the AES weakness 200–1,000 times faster than human cryptography experts and largely developed the AES exploit autonomously; two human researchers then spent about a month verifying the method. Hawk developers reportedly confirmed the attack technique could work. Anthropic researcher Nicholas Carlini said modern AI models are now performing advanced cryptographic research that earlier models did not. While widely deployed encryption systems are not believed to be in immediate danger, observers warn the findings raise long-term concerns for internet security, privacy, and the robustness of future post-quantum standards. The episode highlights how rapidly AI capabilities are advancing in security-sensitive domains.
An AI model finding cryptographic vulnerabilities has broad, cross-industry implications for security, privacy and future encryption standards; this could materially affect how systems (including identity and data protection used in AdTech/MarTech) are designed.
Track Anthropic Signals & Market Shifts in Real-Time
Polaris7 autonomous intelligence agents track regulatory filings, primary sources, executive changes, and deal flow 24/7. Create your free Explorer workspace to monitor these entities.
Key Takeaways & Evidence Grounding
- Claude Mythos reportedly found vulnerabilities in a weakened variant of AES (Advanced Encryption Standard).
- Reporting says Mythos found the AES weakness 200–1,000 times faster than human cryptography experts.
- Mythos largely developed the AES attack autonomously; two human researchers then spent about a month verifying the method.
- Mythos also produced an improved attack against Hawk, a post-quantum candidate; Hawk developers reportedly confirmed the technique could work.
- Anthropic researcher Nicholas Carlini said modern AI models are performing advanced cryptographic research not seen in earlier models.
Connected Companies & Entities
7 Entities mapped“The Anthropic AI Claude Mythos Preview is said to have found weaknesses in a weakened version of the AES encryption algorithm....”
“As the New York Times reported, Claude Mythos Preview identified vulnerabilities in a weakened AES version and was reportedly 200–1,000 time...”
“The article includes external content from TargetVideo GmbH that complements the editorial offering on t3n.de....”
“This article was published on t3n.de under the title 'Claude Mythos: Anthropic-KI knackt Verschlüsselungsstandard – was das bedeutet'....”
Ontology Mapping & Concepts
Related Market Signals & Shifts
Recent verified developments and strategic activity across this market segment.
Anthropic AI Found Vulnerabilities in Encryption Standards
Anthropic's Claude Mythos Preview reportedly identified weaknesses in a weakened version of the AES encryption standard and executed an improved attack on the post-quantum candidate Hawk. According to reporting by the New York Times, Mythos was 200–1,000 times faster at finding the weakness than human cryptography experts had been. Anthropic researcher Nicholas Carlini said model capabilities have advanced rapidly since 2025. Mythos is said to have developed the AES attack almost fully autonomously, after which two human researchers spent nearly a month verifying the method. Hawk's developers have confirmed that the attack approach chosen by Mythos could work, raising questions about long-term implications for cryptographic standards, internet security, and the role of advanced LLMs in vulnerability discovery.
Anthropic Keeps Claude Mythos Private Over Security Risks
Ewor, a Berlin-based startup accelerator positioning itself as a European competitor to Y Combinator, has raised about $70 million from investors to fund its program and equity investments in portfolio companies. Founded and led by Daniel Dippold, Ewor runs an application-driven accelerator that helps early teams hire, raise follow-on funding and reach initial revenues; thousands apply annually. The fund takes equity in participants and aims to scale into a billion-dollar company. Ewor’s portfolio includes fintech Zuba, which uses stablecoins for cross-border transfers. The Ewor team includes experienced founders such as SumUp co-founder Petter Made and Paul H. Müller (known for selling Adjust). Dippold highlighted Europe’s AI and university strengths (ETH Zurich, TU Munich) and noted significant applicant interest from countries like Poland in a Finance-Forward/manager-magazin podcast with editor Carsten Schlenk.
Anthropic Limits Mythos AI Rollout Over Cyberattack Fears
Anthropic released a preview of its frontier model, Mythos, and is deploying it selectively under a new security initiative called Project Glasswing. Twelve partner organizations — including Amazon, Apple, Broadcom, Cisco, CrowdStrike, the Linux Foundation, Microsoft and Palo Alto Networks — will pilot Mythos for defensive cybersecurity work; Anthropic said an additional 40 organizations will gain preview access beyond the partner cohort. Although Mythos was not explicitly trained for security, Anthropic says the model identified “thousands of zero-day vulnerabilities,” many decades old, when scanning first‑party and open‑source code. The rollout follows a prior leak of drafts (the model was previously referenced as “Capybara”) and an accidental exposure of source code tied to a Claude Code release. Anthropic said it is in discussions with federal officials even as it faces legal and supply‑chain disputes with the U.S. government.
Track Real-Time Market Signals & Shifts
Set up custom watchlists to receive automated, evidence-grounded executive digests whenever material signals or shifts occur across your tracked landscape.