Industry
Cloudflare uses AI to map its post-quantum migration
Cloudflare is racing to be quantum-ready by 2029 and is building an internal tool called CryptoLabe to help it find and understand cryptography in its codebase.
YESTERDAY, STILL WORTH KNOWING
Still
Crunchy.
Cloudflare races to quantum-proof its systems while AI models become both hackers and defenders, and hardware makers push the boundaries of power and portability.
Quantum Readiness and AI Governance
Cloudflare is racing to be fully post-quantum ready by 2029, using an internal tool called CryptoLabe to map how cryptography is used across its codebase. The tool uses AI to scan repositories and configuration files, moving beyond simple pattern matching to understand the context of encryption. This helps the company identify protocols that lack a migration plan and ensures its infrastructure can withstand future quantum adversaries. As an infrastructure provider, Cloudflare's move sets a new standard for how large organisations manage the security of their digital foundations.
Separately, Reco has raised $55 million to address the governance of AI agents. The company provides visibility and control over AI workflows, mapping permissions and connections to identify risks. Its solution integrates with major providers like OpenAI and Microsoft Copilot, helping security teams prevent unauthorised access. As AI agents become embedded in business operations, tools like Reco are becoming essential for managing their permissions and enforcing policies.
AI as a Hacker and Defender
Cloudflare has used frontier AI models to stress-test its own Web Application Firewall. The security firm built a system where an LLM acted as a hacker, iterating on payloads to see if the WAF would block them. The tester ran 1,107 attempts across categories like SQL injection and cross-site scripting. While the vast majority were blocked, the few that got through helped Cloudflare create new detections to harden the system. This experiment provides a blueprint for how security teams can use AI to stress-test their own defences.
In a different application of the technology, Google DeepMind has unveiled SynthID Bio, a proof-of-concept that embeds an invisible watermark into AI-generated proteins. The method applies the watermark directly to the amino acid sequence, preserving the biological function while making it possible to identify the content as AI-generated. This addresses the need for research integrity as AI tools become capable of designing novel biological molecules.
Hardware and Gaming Developments
Intel's next-generation Nova Lake platforms have cleared USB and PCIe compliance tests, confirming support for USB4 80Gbps and PCIe 5.0 x16. The company used a program called Qualification by Similarity to certify select products, which will qualify related hardware with limited additional verification. The successful completion of these tests signals Intel is on track for its next desktop and mobile CPU refresh in Q1 2027.
In the gaming sector, Sony Japan has introduced a lottery system for PS5 Pro purchases to curb scalpers. The Sony Store will only accept applications from users with a Japanese PSN account who have logged at least 60 hours of gaming on a PS4 or PS5. Applications open on October 7, with sales to ticket holders starting October 26. Meanwhile, PlayStation Plus subscribers can claim three new games for October: Earth Defense Force: World Brothers 2, F1 25, and Hunt: Showdown 1896.