Follow the latest coverage, related explainers and connected technology stories.
The Linux Foundation’s Akrites initiative is working to validate AI-generated vulnerability reports, merge and improve duplicate reports before sending them to open-source project maintainers. Around 200 seconded engineers are helping keep human review in the loop, as automated reports have become more plentiful but still suffer from inaccuracies and a lack of context.
Microsoft says Defender missed fewer high-severity threats than the other solutions included in the benchmark for the period from May to July 2026, with a clear advantage in post-delivery remediation. The results also show that artificial intelligence is increasing attackers’ ability to impersonate others, and that effective protection requires continuous measurement and multiple layers.
GitHub draws a set of practices for evaluating LLM systems before production from an evaluation experiment involving a language-model-based system designed to reduce false positives in secret scanning. The experiment emphasizes linking metrics to the product decision, simulating the actual operating environment, and analyzing errors, while treating offline evaluation as structured evidence rather than a guarantee of behavior in every production scenario.