Every significant capability gain published in the last 48 hours came from changing the harness around the model instead of the model itself, and none of it shipped with a security evaluation.

Every significant capability gain published in the last 48 hours came from changing the harness around the model instead of the model itself, and none of it shipped with a security evaluation.

Three labs spent this week engineering containment against their own models, and the thing being contained is offensive security capability that arrived faster than any of them planned for.

Five gates went up around the AI stack in forty-eight hours, and the GitHub daily board is quietly voting for everything you can pick up and carry out.

AI now reviews code and attacks it, and only the attacking side gets to iterate against live feedback.

Three separate moves in 48 hours all changed the layer between your app and the model, and not one of them was a model.

Four separate things that were free or open picked up a gate in 72 hours, and the counter-tooling is already climbing the trending charts.

The competition moved off the model and onto the harness, and the plugin ecosystem that formed around DeepSeek Harness in 72 hours is what a platform land grab looks like before anyone calls it one.

Offensive security capability became the thing labs gate releases on this week, and the same week's speed and locality launches make that gate almost impossible to hold.

StateM's reproducible claim is the roughly $15 price rather than the 95.3% score, because Terminal-Bench's published leaderboard subtracts…

Bankruptcy court has become a training-data supply line, and the privacy machinery in the code protects the customers a dead company had,…

Foundry moved agent tool governance into per-request parameters, and Microsoft's own operational checklist says the denylist form of that…

Career-ops's real product is a refusal threshold, and its real risk is that the same agent enforcing the threshold will rewrite the rubric…

Watermarks-remover is the clearest published account of why text watermarking fails as a trust primitive, because its README documents that…

DSH Desktop's own known-limitations section says its auto-updater validates the download container rather than publisher identity, which is…

Computer History is the best-documented agent memory feature anyone has shipped, and its documentation tells you the derived memory files…

Codex resolves which models you may delegate to from a static server-side model catalog rather than from your config, so a documented…