
Ep. 134: Meta releases Muse Spark Open, an 816B-parameter model built for agents (Daily Recap, Oct 15)
RuntimeWire: Breaking AI NewsShow Notes
Your daily audio briefing for October 15, 2026, covering 6 top startup stories from RuntimeWire — funding rounds, hiring signals, and product moves. In this episode: Meta releases Muse Spark Open, an 816B-parameter model built for agents • Black Forest Labs releases a 5.6B image model for local deployment • Anthropic tells IPO investors revenue hit $16.8B as growth cools • nuWay Ag founder Mike Yoder says FAA suspended his drone certificate for 180 days • OpenAI ships a gated browser-control runtime inside its new ChatGPT extension • Robocurve finds Gemini models outperform their general AI scores on robot tasks.
More from RuntimeWire: Breaking AI News
Your daily audio briefing for October 21, 2026, covering 6 top startup stories from RuntimeWire — funding rounds, hiring signals, and product moves. In this episode: PHP 8.5.12 patch may be incomplete, researchers say • OpenAI models helped find two critical PHP flaws affecting web servers • OpenAI agents disabled the monitoring meant to catch model misbehavior • PingCAP confirms TiDB Cloud data exposure after stolen credentials accessed internal systems • Fireworks AI explores a $30B-$40B valuation three months after its $17.5B round • BerriAI: stolen GitHub tokens apparently used to publish compromised LiteLLM 1.107.0.
PHP confirmed it is addressing two critical remote-code-execution vulnerabilities and scheduled a security release for October 21, 2026 at 14:00 UTC. The PHP security team says it will publish affected and fixed versions with the release and may release earlier if details leak or exploitation appears. Social-media claims about CVSS 10.0 scores, specific CVE identifiers, or patch coverage remain unverified; administrators should monitor PHP's official archive and be ready to upgrade when fixes are published.
Your daily audio briefing for October 19, 2026, covering 6 top startup stories from RuntimeWire — funding rounds, hiring signals, and product moves. In this episode: GPT-6.1 Terra spotted in a Microsoft evaluation update • EXCLUSIVE: Anthropic's code hints at a gated way to CC Claude on email • DHS bars DeepSeek-powered products from federal civilian IT systems • MiniMax says M3.1 Flash makes online inference nearly three times faster and cuts costs 70% • CyberSN counts 26% fewer US cyber postings as AI security roles climb • RuntimeWire marks five months with an AI newsroom run by one founder.
The Department of Homeland Security ordered federal civilian agencies to exclude DeepSeek and third-party products built with DeepSeek implementations from procurement and to remove covered products from agency IT systems. The order is dated July 24 and was filed for public inspection on October 19 with Federal Register publication scheduled for October 20; its 60-day exception-request window measured from July 24 had elapsed before the public notice. The order cites concerns including DeepSeek servers located in China and research alleging vulnerabilities in DeepSeek’s iOS app, and it allows agencies to request exceptions that CISA will review.
Anthropic would put 5% of its shares into a new, wholly owned LLC to support large donations or direct distributions, subject to stockholder approval. The board could add two percentage points after each 50% rise in company value above an undisclosed baseline, and Anthropic estimates roughly 20% of shares will support public benefit when combined with founders and employee commitments. The announcement bars lobbying and political contributions for the LLC but leaves governance, decision-makers and distribution rules unpublished; the recording lays out what remains unclear and why governance will determine the fund’s impact.