No themes have been published for this stream yet.
RegLab scanned roughly 3B municipal, 500M state, and 33M federal words of legal code. It surfaced obsolete or unlawful provisions at scale, including dozens of municipalities with language on segregated schools, poll taxes, or male-only voting eligibility; identifying the text does not itself repeal it.
The essay tracks the shift from model-graded rewards and synthetic data to self-generated curricula, research loops, environments, and simulated users. Its useful synthesis is that each step became load-bearing only when verification improved; the examples span years and are not one controlled comparison.
Claude Security scans now run on Claude Mythos 5, available today in public beta for all Claude Enterprise customers. Put our most capable security model to work on your codebase, no separate model access needed.
discussion2 selected replies
The public beta lets Enterprise customers run Mythos 5-backed repository scans without direct model access. Anthropic says findings include CWE, confidence, severity, and a suggested fix, and that scans use standard token billing under the existing plan.
We analyzed GLM-5.3 and Claude Fable 5 on DeepSWE. GLM-5.3 MATCHED Fable 5 on single-shot solve rate at about one-fifth the cost, then pulled ahead with multiple attempts. The economics get interesting fast when retries are cheap 👇
Together reports that GLM-5.3 matched Claude Fable 5’s single-shot DeepSWE solve rate at about one-fifth the cost, then led with multiple attempts. The post omits the full sample and harness settings, so this remains a provider’s cost-normalized result.
Ox Alpha is the fastest growing model in Cline’s history. It’s already driving 8% of all inference through Cline in just 48 hours.
Cline says Ox Alpha became its fastest-growing model and reached 8% of platform inference within 48 hours. The first-party usage metric measures adoption, not quality, and Cline does not identify the model’s maker.
5 posts · 86% confidence
AI is now detectable in software-coder employment without explaining the wider hiring slowdown. A Fed counterfactual identifies a likely post-ChatGPT coder shock; Richmond Fed finds workers less exposed to AI still account for roughly half of the aggregate fall in job-finding since mid-2023.
The three-day tariff pause ended without a deal: the United States imposed 50% tariffs on roughly $28B of Canadian goods after Canada rejected revised terms. Canada will answer dollar for dollar on selected sectors after Labour Day, converting implementation risk into a bilateral cost shock.
Tonight, Canada declined to finalize the trade deal under the terms agreed earlier this week. Despite the U.S. offer to Canada to receive the best treatment of any major exporter to our market, new demands and walk backs of other commitments by Canada have upended the careful…
Tonight, Canada declined to finalize the trade deal under the terms agreed earlier this week. Despite the U.S. offer to Canada to receive the best treatment of any major exporter to our market, new demands and walk backs of other commitments by Canada have upended the careful balance reached in the past days. In addition, Canada is continuing to maintain its prolonged retaliation against the United States, including, among other things, flat-out prohibitions on certain American goods and services. For decades, Canada has enjoyed the most favorable access to the U.S. market of any country. And from the beginning of President Trump’s trade program, Canada has continued to enjoy the best treatment in the world, even after – like China – retaliating against the United States. This week, the United States agreed to provide even better treatment to Canada, offering significant tariff reductions on steel, aluminum, autos, and lumber. The U.S. offer was also forward looking, and included a historic economic and national security partnership to cooperate on export controls, combat transshipment, enhance digital trade, and align certain external tariffs. The offer would have led to supply chain coordination on aerospace, complementary actions to address unfair trade practices, critical minerals cooperation, increased enforcement against imports produced with forced labor, and the announcement of formal U.S.-Mexico-Canada Agreement (USMCA) negotiations. This is a missed opportunity for Canada to partner with the United States, which is the fastest growing economy in the G7.
Bankers told prospective investors that Anthropic could seek more than $100B in an IPO, in discussions that would imply a valuation near $2T. This is banker marketing, not a filed deal or a company decision, but it sets an unusually explicit financing scale.
No single-source signals cleared the bar today.
3 posts · 88% confidence
Point Claude Security at a GitHub repo and Mythos scans for vulnerabilities tracing data across files and reasoning about how components interact. Each finding comes back with a CWE category, confidence and severity ratings, and a suggested fix.
Suggested patches open in Claude Code on the web, using the models your team already uses. These updates are part of our work to give defenders greater access to Mythos results without requiring direct access to the model: the model runs behind the scan and returns findings only. Scans are billed as standard token usage under your existing plan.