All Stories — Page 49
Every story published on RuntimeWire, newest first.
- Snorkel AI raises $350M after turning its software into a data service
The Stanford spinout says its data-service pivot lifted annualized revenue run-rate from $20M to $350M in a year.
- GNU Radio joins OCUDU, giving open-source RAN a physical-layer workbench
Eric Blossom's 25-year-old SDR project enters OCUDU as the Linux Foundation initiative builds a shared stack for 5G and 6G research.
- Cisco Talos finds malware taking orders from a four-model committee
Ryan Fetterman's open-source CAIRN surfaced CLOSEDQUORUM, a Windows implant whose AI voting loop is confirmed, though live attacks are not.
- DeepSeek invited to make a statement at UN Security Council AI meeting
Reuters says DeepSeek and other Chinese AI developers were invited to make statements at the September 23rd meeting, but DeepSeek's attendance, delegate and remarks remain unconfirmed.
- IAB Tech Lab proposes OpenProposal to automate the ad-buying RFP slog
OpenProposal connects buyer briefs, seller proposals and buy commitments, with public comment open through October 22nd.
- Verda raises $189M for the GPU cloud investors once called too capital intensive
Emergence Capital led the Series B; Verda says it is worth at least $1B and plans more than 250 MW of operations in 2027.
- Tripo AI ships P2.0 with native quad meshes for production pipelines
Tripo AI's model adds regional mesh editing and four variants per prompt, 20 days after the company announced 3 billion yuan in funding.
- Tencent ships Hy Image 3.5 Preview with multi-turn editing at $0.024
The preview accepts text and reference images, keeps editing context across turns, and undercuts Hy Image 3.0's listed API price by 25%.
- Instinct user says the AI assistant surfaced a stranger's financial document
Screenshots show Instinct blaming a crossed chat, but its own admission cannot prove whether the episode was a leak or a hallucination.
- South Korean chip stocks rally as Meta's Muse joins AI optimism
Samsung and SK Hynix rose on September 22nd as broader AI optimism, lower oil prices and softer US yields lifted Seoul stocks. Bloomberg linked Muse to investor sentiment, not disclosed chip orders.
- GPT-6 Astra reaches 53% in Rails coding benchmark as Gemini regresses
GPT-6 Astra reached 53% accuracy, Gemini regressed, and a previously scoreless OpenAI model suddenly completed 16 runs.
- Microsoft adds Teams message snippets to Defender threat hunting
The MessageContents table will expose content and metadata to authorized analysts, forcing administrators to revisit who can read collaboration data.
- RunAnywhere launches Wally, says its open-model inference outruns larger providers
The YC W26 company is extending its on-device stack into hosted inference, claiming 380 tokens per second on GLM-5.3 Flash.
- Meta's Muse flaw lets Mac malware reach linked iPhones, researcher says
Patrick Wardle's proof of concept redirected Muse traffic, exposed account access and remotely queried an iPhone's location and nearby Bluetooth devices.
- Wafer claims a 12T monthly token run rate and keeps hiring engineers
The San Francisco inference startup is offering up to $300K for engineers who will own kernels, clusters and customer workloads.
- Bond recasts its AI chief of staff as a to-do list that works back
The YC-backed product connects Slack, email, calendars and project tools, then ranks commitments and prepares the next action.
- Investigation finds more than 110 deaths within range of Anduril border towers
The findings test Palmer Luckey and Brian Schimpf's bet that autonomous detection can move government operators fast enough to save lives.
- Xiaomi open-sources MiMo-V2.6 and the RL machinery behind it
Xiaomi released two omnimodal model repositories, reinforcement-learning resources and a technical report while retaining MiMo-V2.5 API prices.
- Meta plans a 1 Pbps transatlantic cable using two-core fiber
Petal is due in 2029, linking the US and France with twice the capacity of Meta's 0.5 Pbps Anjana system.
- Exclusive: Vibe coding is coming to the enterprise. UST rewrites or rejects 40% of the output
UST Chief Solutions Officer Kailash Attal tells RuntimeWire how Codon brings plain-English building to five major enterprise platforms while keeping every deployment behind approval gates, audit trails and mandatory human review.
- A vibe-coded timeline packs 42 years of Mac buttons into one control
Designer Cole Derochie used Astra and a Figma reference to build 11 interactive styles, then corrected the chronology in public.
- Linear reworks CI as coding agents make validation the expensive part
Tuomas Artman's team cut runner time per test roughly in half while Linear's test suite nearly quadrupled in 2026, according to Linear.
- Aikido ships Altar-1, a 328 GB GLM-5.3 prune for on-prem pentesting
The open-weight release runs on four H200 GPUs and powers Aikido Machine, keeping code and pentest data inside customer networks.
- Kyutai releases speech-native math models that skip the text LLM
The released GLM-4-Voice derivatives score up to 77.1% on GSM8K, though larger cascaded systems still lead the benchmark.