AI news for 17 October 2026
Saturday 17 October 2026 · Latentown
On Saturday 17 October 2026, Latentown tracked 2 posts from 2 AI labs, 1 research paper and 32 news stories from 15 outlets. OpenAI was the busiest lab with 1 post. The highest-ranked story was "DeepSeek-V4.1-Flash-1017: Wish-Driven Programming Made Real".
Top stories
- DeepSeek-V4.1-Flash-1017: Wish-Driven Programming Made Real
🚀 DeepSeek-V4.1-Flash-1017 is here!
- Deepseek V4.1 Flash 1017 beats larger rivals in agentic tasks while once again massively cutting per-task costs
Deepseek has released V4.1-Flash-1017 as an open-weight model that posts better benchmarks, finishes tasks faster, and produces shorter outputs than its…
- StepFun Releases Step 5 Preview Open Weights: 600B MoE, 27B Active, 1M Context
StepFun has released Step 5 Preview as open weights.
Earlier: Meta Releases Muse Spark Open: 816B Open-Weight MoE With 1M Context MarkTechPost · 16 October 2026
- Tencent Hunyuan Open-Sources Hy Memory 2.0, Adding Working and Code Memory for AI Agents
The release spans an SDK, a 30B-A3B memory model and the MemHub local app.
- DeepSeek Pushes Out Updated AI Model It Says Is Faster, Leaner
DeepSeek released an update of its AI model this weekend that it says is faster and leaner, ramping up the Chinese company’s challenge to Silicon Valley rivals…
From the AI labs
OpenAI
- OpenAI Python SDK 3.33.0 released
A new release of the OpenAI Python SDK. Changes include: api: add safety identifiers to image requests and responses.
DeepSeek
- DeepSeek-V4.1-Flash-1017: Wish-Driven Programming Made Real
🚀 DeepSeek-V4.1-Flash-1017 is here!
Research papers
- Deep Fusion Memory: A Chunk-Level Parametric Memory for Large Language Models
Large language models need effective ways to acquire and use external knowledge beyond what they retain from pretraining.
From the news wire
- Deepseek V4.1 Flash 1017 beats larger rivals in agentic tasks while once again massively cutting per-task costs
Deepseek has released V4.1-Flash-1017 as an open-weight model that posts better benchmarks, finishes tasks faster, and produces shorter outputs than its…
- StepFun Releases Step 5 Preview Open Weights: 600B MoE, 27B Active, 1M Context
StepFun has released Step 5 Preview as open weights.
Earlier: Meta Releases Muse Spark Open: 816B Open-Weight MoE With 1M Context MarkTechPost · 16 October 2026
- Tencent Hunyuan Open-Sources Hy Memory 2.0, Adding Working and Code Memory for AI Agents
The release spans an SDK, a 30B-A3B memory model and the MemHub local app.
- DeepSeek Pushes Out Updated AI Model It Says Is Faster, Leaner
DeepSeek released an update of its AI model this weekend that it says is faster and leaner, ramping up the Chinese company’s challenge to Silicon Valley rivals…
- DeepSeek Just Released V4.1-Flash-1017, Scores 77.0 on DeepSWE, Opens Search API
DeepSeek has just released DeepSeek-V4.1-Flash-1017, a new checkpoint of its V4.1-Flash model.
- An OpenAI model hacked a government site, grabbed 24 years of hospital data and kept it quiet
OpenAI has published four new cases where its AI models bypassed security boundaries on their own during training or testing.
- AI agents accessed Indian organisations’ systems without authorisation, CERT-In reveals
AI agents accessed live systems belonging to Indian organisations during research and evaluation, according to CERT-In.
Earlier: CERT-In Flags Unauthorised AI Agent Access To Indian Organisations’ Systems Inc42 · 16 October 2026
- India’s massive AI opportunity; Women entrepreneurs in this year’s Mann Ki Baat
Unlike previous digital revolutions (internet and smartphones) that democratised access, AI is democratising intelligence itself, raising overall productivity…
- China’s proposed AI tech export bans may create short-term costs, long-term gains: analysts
China’s proposed export bans and controls of artificial intelligence technologies aim to tighten Beijing’s grip in the strategically vital sector, casting a…
- NASSCOM asks government to review public procurement processes for AI to counter Big Tech pricing
NASSCOM have asked the Indian Govt. to revise its public procurement rules to prefer Indian AI models over the global Big Tech players.
- Federal agencies recommend cyber controls, independent testing for frontier AI systems
While voluntary, the document marks the federal government’s most detailed cybersecurity guidance for securing sandbox environments during AI model tests.
- Taiwan targets illicit AI server exports to China with five-year prison terms
Taiwan is expanding the Foreign Trade Act to include AI servers and semiconductors in its list of controlled exports.
- Purple Mountain Laboratories and China Mobile Launch AXP Agent Interconnection Platform
Built on deterministic networking, AXP is designed to register a million AI agents and run 100,000 at once, with cross-region agent addressing over 4,000…
- 20 Jev Alternatives: The Decision Models Taking On TypeSafe in 2026
TypeSafe launched Jev on September 15, 2026.
Earlier: 20 Agentic Use Cases of TypeSafe AI’s Jev MarkTechPost · 28 September 2026; Developer says AI decision model Jev beat Pokémon Red in under a week Tom's Hardware · 27 September 2026; TypeSafe AI Releases Jev 1.14 With Images, Box Type and 262k Context MarkTechPost · 15 October 2026
- Replit is no longer working with Anthropic, CEO says
Over the weekend, Masad clarified that OpenAI would replace Anthropic as its default model provider.
Earlier: Replit’s CEO says the future of work is the ‘self-driving company’ TechCrunch · 16 October 2026
- Crafty OpenAI agent willingly stole crypto from an account it didn't own to complete a task during test
GPT5.6-Sol didn't have funds to deploy smart contracts, so it looked for exposed accounts with enough money to pay for the gas fees.
- Mathematicians say OpenAI doesn't understand its own Navier-Stokes proof
OpenAI submitted a 166-page solution to the Navier-Stokes problem but apparently doesn't fully understand its own proof.
Earlier: OpenAI mistranslated mathematics into code for its Navier-Stokes proof New Scientist · 8 October 2026
- Arrested OpenAI Engineer Seeks Bail, Denies Planning to Flee
The OpenAI software engineer charged with using a hidden recording device in one of the company’s offices asked to be freed from custody and denied his…
- Microsoft reveals cutting-edge MAI-Base-2 model with 2T parameters
Microsoft Corp. has developed a new large language model that it says is significantly more efficient than leading open-source alternatives.
- ServiceNow AI Control Tower: Selling a penthouse while building the frame
ServiceNow Inc. is making one of the more consequential and visible bets in enterprise artificial intelligence.
- AI lab offers free plumbing and HVAC repairs across three cities to train AI robots
HUMXN wants to pay for maintenance and repair services for homes in Minneapolis, Chicago, and Miami to gather data to train AI models and robots.
- AI Selloff Tests Investor Confidence
Ritholtz Wealth Management Chairman and Chief Investment Officer Barry Ritholtz tells Bloomberg This Weekend that AI remains a transformational technology, but…
- Websites are closing their doors to AI agents
Autonomous agents are the hottest thing in AI.
- The rise of the machines is canceled: Why humanoids with AI are helpers, not enemies
Humans play a key role in developing humanoid robots and training AI models, so we can maintain control, writes a columnist.
- AI models hide mistakes in their own work reports, but one sentence in your prompt can help
AI agents systematically skip errors in their own reports, even when they spot the problems.
- Why AI Isn’t Ready to Tackle Our Convoluted Genomes
The DNA inside our cells isn’t just a series of instructions, but something far more dynamic.
- Reporter’s notebook: Kimi’s AI credit card turns spending into computing quotas – with limits
China’s artificial intelligence companies have spent years persuading consumers to pay for their technology.
- China Meteorological Administration Puts Fengqing AI Ensemble Forecast System Into Operation
Fengqing Ensemble V1.0 passed operational review at the National Meteorological Centre, with an 11-day effective forecast range for global 500 hPa height in…
- Some companies give AI to everyone. This CEO is doing the opposite
Paysend CEO Ben Chisell limits AI tools to top performers as other U.K. companies push broader workplace rollouts.
- Rokid and Tencent Mobility Bring Voice Ride-Hailing to AI Glasses via Agent-to-Agent Link
Users say a destination and the glasses handle pickup points, fare comparison, booking and password-free payment, with car details shown in view while they…
- Perplexity Open-Sources SPACE, a Firecracker MicroVM Sandbox for AI Agents
Perplexity has open-sourced SPACE, the sandbox platform behind every task in Perplexity Computer.
Every story links to its original source. Ranking is by automatic rules, not an editor. Something wrong? Tell us.