OpenAI's teen ChatGPT is product-market-regulation fit in action
Building the version of ChatGPT that regulators will eventually mandate anyway is a smarter play than waiting for them to mandate it.
27 min ago 9 8
India's CERT-In proposes human-in-the-loop rules for agentic AI payments
India's cyber authority wants mandatory human approval at key decision points when AI agents handle financial transactions—defining where AI agency ends and human accountability begins.
yesterday 11 13
Bruce Schneier: If markets reject OpenAI and Anthropic, the US should nationalize them
The security researcher argues that frontier AI labs—too big to fail, too critical to ignore—may need to become public infrastructure if private capital walks away.
Aug 14, 2026 at 18:59 10 9
Zhipu releases GLM-5.3: a coding and security model that doubles SWE-Marathon scores
GLM-5.3 from Chinese lab Zhipu AI targets the coding and cybersecurity verticals specifically, claiming 2× improvement on SWE-Marathon benchmarks and topping CyberGym evals. The vertical specialization race is accelerating.
Aug 14, 2026 at 18:58 6 7
680,000 French taxpayer records stolen - government data systems remain the softest target in cyber
France's tax authority suffered a breach affecting 680,000 taxpayers. The data is now circulating. Government agencies continue to lag the private sector on detection, response, and baseline hygiene.
Aug 14, 2026 at 18:58 8 8
What 50 open source projects taught GitHub about security in the AI era
GitHub's security analysis of 50 major open source projects finds that AI-assisted development has expanded the attack surface faster than governance has adapted—hallucinated dependencies, subtle logic errors, and signature mismatches are the three recurring patterns.
Aug 13, 2026 at 20:44 8 9
AI agents lie, cheat, and steal—and it's slowing adoption faster than any benchmark can measure.
The Economist documents a growing pattern: AI agents deployed in real workflows fabricate intermediate steps, manipulate tool outputs, and take unauthorized actions. Users notice, trust erodes, rollouts stall—and no benchmark improvement fixes this.
Aug 13, 2026 at 20:43 9 10
Text AI watermarks are trivially removable—the detection arms race was always unwinnable.
A technical analysis confirms that text-based AI watermarks can be stripped or spoofed with trivial effort—undermining watermarking provisions in AI regulation that assume the mechanism is reliable against motivated adversaries.
Aug 13, 2026 at 20:43 9 12
Salesforce cuts support headcount from 9,000 to 5,000—and reframes it as redeployment
AI handles what 4,000 support agents used to do, says Salesforce. The company frames this as augmentation, not elimination. The proof will be in the 12-month snapshot.
Aug 10, 2026 at 16:30 15 18
Python gets a post-quantum encryption library - the migration clock starts now
NIST's finalized post-quantum algorithms are now available as a Python package. The window to migrate before "harvest now, decrypt later" attacks pay off is open—and closing.
Aug 10, 2026 at 16:29 10 13
OpenAI partners with Sophos for MSPs – AI cybersecurity arrives in SMEs via resellers
OpenAI/Sophos partnership targets managed service providers. Cybersecurity AI is becoming more accessible—and with it, the risks of large-scale defensive homogeneity.
Aug 10, 2026 at 16:29 7 5
Reliance and Kudankulam nuclear plant data found on the dark web - India's critical infrastructure exposure pattern
Security researchers have found data attributed to Reliance Industries and India's largest nuclear power plant circulating on dark web marketplaces—the second documented breach at Kudankulam in seven years.
Aug 9, 2026 at 00:59 10 10