Privacy and Cybersecurity Roundup Covers New Rules and Project Glasswing
Summary
This roundup covers EU and national privacy and security developments, UK guidance on agentic AI risks, new US privacy laws, FTC priorities, and Anthropic’s Project Glasswing initiative to use AI to find vulnerabilities in critical software.
Key points
- The EDPB’s 2025 report highlights the Helsinki Statement, new GDPR guidance, 414 cross-border cases, 572 final decisions, and approximately €1.15 billion in fines.
- Germany’s BSI updated its C5 cloud security catalogue. The BSI does not certify providers; customers are expected to review reports and conduct their own risk analysis.
- CNIL published HR data retention guidance and set 2026 inspection priorities covering recruitment, the electoral register, and sports federations, including scrutiny of security and data retention.
- A UK DRCF foresight paper says existing laws continue to apply to agentic AI and flags risks including prompt injection and privilege escalation, alongside expectations for oversight, logging, and risk-based controls.
- Alabama passed a consumer privacy bill that would require reasonable security safeguards; Kentucky passed an amendment requiring consent for collection and use of smart-device ACR data.
- The FTC’s 2026–2030 plan includes data security enforcement, children’s online protection, and agency modernization, including zero-trust cybersecurity architecture.
- Anthropic announced Project Glasswing to use its unreleased Claude Mythos Preview for security testing. Anthropic says early testing identified thousands of high-severity vulnerabilities.
Article Details
- Event Type
- Anthropic announced Project Glasswing, a multi-industry initiative using frontier AI to identify vulnerabilities and improve the security of critical software.
- Impact
- According to Anthropic's announcement, its unreleased Claude Mythos Preview model can identify and exploit software vulnerabilities, reportedly outperforming most human experts in this area. Early testing reportedly identified thousands of high-severity vulnerabilities across major operating systems and web browsers. More than 40 organizations maintaining critical infrastructure were granted access for security testing, with up to $100 million in usage credits and $4 million in direct funding for open-source security initiatives. These capabilities and findings are reported by Anthropic.
Vendors
Amazon Web ServicesAI models. The project brings together major technology and infrastructure actors, including Amazon Web Services, Apple, Cisco, Google, Microsoft, NVIDIA, and others, alongside open-source organizations andAnthropicAnthropic Launches “Project Glasswing” to Secure Critical Software Using Frontier AIAppleThe project brings together major technology and infrastructure actors, including Amazon Web Services, Apple, Cisco, Google, Microsoft, NVIDIA, and others, alongside open-source organizations and financialCiscoproject brings together major technology and infrastructure actors, including Amazon Web Services, Apple, Cisco, Google, Microsoft, NVIDIA, and others, alongside open-source organizations and financial institutions.Googlebrings together major technology and infrastructure actors, including Amazon Web Services, Apple, Cisco, Google, Microsoft, NVIDIA, and others, alongside open-source organizations and financial institutions.Microsoftmajor technology and infrastructure actors, including Amazon Web Services, Apple, Cisco, Google, Microsoft, NVIDIA, and others, alongside open-source organizations and financial institutions.Nvidiatechnology and infrastructure actors, including Amazon Web Services, Apple, Cisco, Google, Microsoft, NVIDIA, and others, alongside open-source organizations and financial institutions.
Products
Countries
Germany🇩🇪 Germany: BSI Updates C5 Cloud Security CatalogueUnited Kingdom🇬🇧UK DRCF Publishes Foresight Paper on Agentic AIUnited StatesThe Act introduces a comprehensive consumer privacy framework, making Alabama the 21st U.S. state to adopt such legislation. It largely follows the “Virginia model,” but includes several structural deviations.
Industries
Critical infrastructuredirect funding to open-source security initiatives. Over 40 additional organizations maintaining critical infrastructure have been granted access to the model for security testing purposes.SoftwareAnthropic Launches “Project Glasswing” to Secure Critical Software Using Frontier AI