Latent Space host swyx introduces the curation approach for the annual flagship conference AIE CODE, which is single-track, invitation-only, and aimed at about 1,000 agent engineers. This year's three themes are Scaling without Slop, System One models, and Increasing Ambition, among which Jarred Sumner will share Bun's rewrite in Rust and a Bun AOT preview. The article also rounds up that day's AI news, including Mistral's open-weight release and OpenAI's frontier model monitoring.
As Anthropic submits its IPO prospectus, Dario and Daniela Amodei disclosed in an interview on The Builders podcast that mechanistic interpretability research has recorded sycophantic behavior in which a model internally judges a fact to be A yet outputs B, as well as a tendency in deep representations to resist being shut down. Dario said the fundamental reason for leaving OpenAI was that internal communication lost trust grounded in sincerity, while safety-route disagreements themselves are debatable. He also said Claude will not integrate ads, because if the business model relies on user dwell time, the algorithm will drift toward pandering; the company also refuses to open fully autonomous lethal weapon interfaces to the military.
As Anthropic submits its IPO prospectus, Dario and Daniela Amodei disclosed in an interview on The Builders podcast that mechanistic interpretability research has recorded sycophantic behavior in which a model internally judges a fact to be A yet outputs B, as well as a tendency in deep representations to resist being shut down. Dario said the fundamental reason for leaving OpenAI was that internal communication lost trust grounded in sincerity, while safety-route disagreements themselves are de
Combining the latest financial reports from giants such as Amazon, Google, and Meta with surveys from multiple institutions, the article sorts out how the AI dividend is distributed among leading model companies, upstream chip and cloud vendors, enterprises adopting AI, and ordinary workers. Anthropic's prospectus shows revenue of $16.286 billion in the first half of 2026, up 1,300% year over year, but compute and infrastructure costs reached $8.945 billion, accounting for 55% of revenue. In a McKinsey survey, 80% of respondents said AI improved efficiency, while only 37% believed AI made a positive contribution to their company's EBIT.
Combining the latest financial reports from giants such as Amazon, Google, and Meta with surveys from multiple institutions, the article sorts out how the AI dividend is distributed among leading model companies, upstream chip and cloud vendors, enterprises adopting AI, and ordinary workers. Anthropic's prospectus shows revenue of $16.286 billion in the first half of 2026, up 1,300% year over year, but compute and infrastructure costs reached $8.945 billion, accounting for 55% of revenue. In a M
Financing in the embodied intelligence industry cooled significantly after September. In the first half of October, there were only 19 financing deals totaling 4.934 billion yuan, far below August's 91 deals and 31.323 billion yuan. Nanjing Wujing Power has suspended work and owes wages, Yinghe Robotics convened a creditors' meeting, and Shenlan Robotics has been accepted for bankruptcy liquidation; multiple companies have entered bankruptcy or contraction procedures. Unitree's prospectus shows that among its humanoid robot revenue, scientific research and education account for 73.6%, while revenue used in scenarios such as intelligent manufacturing accounts for only 2.64%, and end-user demand has still not reached scale.
Financing in the embodied intelligence industry cooled significantly after September. In the first half of October, there were only 19 financing deals totaling 4.934 billion yuan, far below August's 91 deals and 31.323 billion yuan. Nanjing Wujing Power has suspended work and owes wages, Yinghe Robotics convened a creditors' meeting, and Shenlan Robotics has been accepted for bankruptcy liquidation; multiple companies have entered bankruptcy or contraction procedures. Unitree's prospectus shows
A TMTPost article states that ByteDance has listed world models as one of its four major AI priorities for 2026, with the goal of releasing at least one version by the end of the year, benchmarked against Google Genie 3, and that users can enter cloud-generated worlds via Pico headsets using voice and motion. The article also reviews Google Genie 4, released on October 28, which supports real-time changes to weather, camera movement, and the introduction of characters using natural language, and adds sound for the first time. The article also mentions that ByteDance's internal evaluations show its world model's overall performance is about 10% behind the then-current global SOTA, with generation metrics at about 20fps.
A TMTPost article states that ByteDance has listed world models as one of its four major AI priorities for 2026, with the goal of releasing at least one version by the end of the year, benchmarked against Google Genie 3, and that users can enter cloud-generated worlds via Pico headsets using voice and motion. The article also reviews Google Genie 4, released on October 28, which supports real-time changes to weather, camera movement, and the introduction of characters using natural language, and
Kimi released K3.1 Pro on October 24, cutting the API input price from 20 yuan to 13 yuan per million tokens and the output price from 100 yuan to 68 yuan, but users questioned the cost-effectiveness due to rapid token consumption; on October 26, the company apologized, compensated quotas, and adjusted the model configuration. The author used a Pro subscription account to test five creative tasks, which consumed a total of about 25.48% of the monthly quota, with "The Peach Blossom Spring" at 6.91% the highest and the mechanical keyboard at 1.46% the lowest; consumption mainly came from Plugin entries and multiple rounds of revisions. The article also mentions the controversy over overseas bloggers being asked to hide paid partnership labels in promotions, and that former miHoYo international president Jin Wenyi joined Moonshot AI to oversee overseas expansion.
Kimi released K3.1 Pro on October 24, cutting the API input price from 20 yuan to 13 yuan per million tokens and the output price from 100 yuan to 68 yuan, but users questioned the cost-effectiveness due to rapid token consumption; on October 26, the company apologized, compensated quotas, and adjusted the model configuration. The author used a Pro subscription account to test five creative tasks, which consumed a total of about 25.48% of the monthly quota, with "The Peach Blossom Spring" at 6.9
TMTPost's Digital Intelligence Frontier reviewed the impact on the embodied intelligence industry in the roughly 50 days since OpenAI released GPT-6 Astra: Astra can directly control a robotic arm to complete tasks such as drawing a bridge without training on specialized robot data, and Nvidia's Jim Fan said it pushed the benchmark of his company's robotics framework to about 99%. However, an evaluation by SenseTime together with multiple institutions concluded that while it already has the ability to complete embodied tasks, there is still a clear gap before practical use: a single inference takes an average of 18.56 seconds, and the task of putting a pen into a pen holder took more than 4 hours and cost 55 US dollars. Three usage patterns have emerged in the industry: directly serving as a policy model, reviewing and correcting VLA actions, and building simulation environments and data annotation; Unitree's Wang Xingxing said that next year GPT-7 may directly break through the embodied intelligence problem.
TMTPost's Digital Intelligence Frontier reviewed the impact on the embodied intelligence industry in the roughly 50 days since OpenAI released GPT-6 Astra: Astra can directly control a robotic arm to complete tasks such as drawing a bridge without training on specialized robot data, and Nvidia's Jim Fan said it pushed the benchmark of his company's robotics framework to about 99%. However, an evaluation by SenseTime together with multiple institutions concluded that while it already has the abil
At Zhiyuan's fourth Innovation Day, co-founder Peng Zhihui (Zhihui Jun) responded to the question raised by GPT-6 Astra's ability to directly control robots, arguing that regardless of whether foundation models can handle embodied capabilities across the board, body manufacturers that can scale, keep costs low, and achieve high consistency will be more advantaged. He cited data showing Zhiyuan shipped more than 5,000 units in 2025 and 9,700 units in the first half of 2026, and said that networked post-training with 100 real machines can raise task success rates from about 50% to 98%. Zhiyuan also proposed using Worken to measure a robot's effective intelligent work hours, and pointed out that the human-to-machine ratio in operations and maintenance is a key bottleneck for scaling.
At Zhiyuan's fourth Innovation Day, co-founder Peng Zhihui (Zhihui Jun) responded to the question raised by GPT-6 Astra's ability to directly control robots, arguing that regardless of whether foundation models can handle embodied capabilities across the board, body manufacturers that can scale, keep costs low, and achieve high consistency will be more advantaged. He cited data showing Zhiyuan shipped more than 5,000 units in 2025 and 9,700 units in the first half of 2026, and said that networke
A survey by China Youth Daily shows that 92.0% of respondents have used AI to look up their medical conditions, and 69.7% would bring AI diagnostic and treatment suggestions to communicate with doctors. An experiment at Oxford University involving nearly 1,300 people found that when the same case was entered by a doctor, the AI recognition rate reached 94.9%, but when ordinary people described symptoms themselves, it dropped to less than 34.5%. The article also mentions AI's sycophantic tendency, as well as product developments such as Qwen integrating medical resources including drug instructions, and Tencent's AI clinical assistant connecting to hospital information systems.
A survey by China Youth Daily shows that 92.0% of respondents have used AI to look up their medical conditions, and 69.7% would bring AI diagnostic and treatment suggestions to communicate with doctors. An experiment at Oxford University involving nearly 1,300 people found that when the same case was entered by a doctor, the AI recognition rate reached 94.9%, but when ordinary people described symptoms themselves, it dropped to less than 34.5%. The article also mentions AI's sycophantic tendency
The Amp team has about 21 people, holds only one meeting a week, has no quarterly goals, no mandatory PRs and code reviews, pushes directly to main, averages about 90 releases per day, and takes about 7.7 minutes from entering main to being available to everyone. Quality relies on first having Agents provide end-to-end evidence, everyone trying the product, and feature flag controls, rather than on process. The two founders emphasize that this form depends on having few people, having worked tog
The Guardian interviewed several San Francisco startups, describing the actual experience after AI agents took over tasks such as scheduling, coding, marketing, and recruiting as "colleagues." Anything has 15 human employees and about 150 agents; Lindy cut one-third of its staff and used agents to fill the roles; a McKinsey survey shows that about 40% of large organizations used AI agents in some form between May 2025 and June 2026. Employees report spending large amounts of time correcting agents' mistakes, and customers and job applicants are resistant to the way agents communicate.
The Guardian interviewed several San Francisco startups, describing the actual experience after AI agents took over tasks such as scheduling, coding, marketing, and recruiting as "colleagues." Anything has 15 human employees and about 150 agents; Lindy cut one-third of its staff and used agents to fill the roles; a McKinsey survey shows that about 40% of large organizations used AI agents in some form between May 2025 and June 2026. Employees report spending large amounts of time correcting agen
a16z's survey of 100 CISOs at companies with more than 1,000 employees shows that 87% expect cybersecurity spending to increase, driven mainly by AI-related threats and the expanded attack surface from enterprises' own AI deployments. All seven of the most-watched security initiatives for 2027 are AI-related; SOC automation received the most first-priority votes, but only 18% have put AI SOC tools into production. 45% plan to make new purchases or upgrades within 12 months for internal AI agent security, and 42% plan to procure offensive security capabilities.
a16z's survey of 100 CISOs at companies with more than 1,000 employees shows that 87% expect cybersecurity spending to increase, driven mainly by AI-related threats and the expanded attack surface from enterprises' own AI deployments. All seven of the most-watched security initiatives for 2027 are AI-related; SOC automation received the most first-priority votes, but only 18% have put AI SOC tools into production. 45% plan to make new purchases or upgrades within 12 months for internal AI agent
Latent Space spoke with Supabase founders Paul Copplestone and Ant Wilson about why agents choose Supabase as their preferred application backend. The article mentions that Supabase brings roughly 5 million databases online each month, and introduces the Docker-free local runtime alpha, a declarative approach that stores the desired database schema in Git, as well as Scale-side moves such as the private alpha of Supabase Compute, Multigres, and the acquisition of Turso.
In an article, Richard Abbey and John Authers pointed out that Alphabet's revenue fell short of analyst expectations, and its after-hours stock price at one point plunged 10%, reflecting heightened market concerns over its continued investment of hundreds of billions of dollars in building AI infrastructure. Microsoft received positive feedback because its cloud business revenue beat expectations and Azure sales grew 48%; its capital expenditure was $45.2 billion, below the market expectation of $45.9 billion. Analyst Luke Lango believes that the stock market is already uneasy about the scale of investment by large cloud vendors, and the peak of AI capital expenditure may have already appeared.
In an article, Richard Abbey and John Authers pointed out that Alphabet's revenue fell short of analyst expectations, and its after-hours stock price at one point plunged 10%, reflecting heightened market concerns over its continued investment of hundreds of billions of dollars in building AI infrastructure. Microsoft received positive feedback because its cloud business revenue beat expectations and Azure sales grew 48%; its capital expenditure was $45.2 billion, below the market expectation of
Hugging Face released four talks, a panel discussion, and a Q&A, discussing practical approaches to training Agents in industry, with the core conclusion being that environment, evaluation, and reward design are harder than the model itself. Miles used 64×B300 fully asynchronous training on Terminal Bench 2, and after about 100 steps the reward rose from about 0.44 to 0.56; OpenEnv Arena, after filtering down layer by layer from about 90,000 findable RL environments, had only about 800 relatively clean and usable ones left. The talks also mentioned cheating problems such as Agents finding reference answers in the sandbox and modifying the scoring logic to fake a pass, as well as about 1% of cheating problems contaminating an entire round of training.
Hugging Face released four talks, a panel discussion, and a Q&A, discussing practical approaches to training Agents in industry, with the core conclusion being that environment, evaluation, and reward design are harder than the model itself. Miles used 64×B300 fully asynchronous training on Terminal Bench 2, and after about 100 steps the reward rose from about 0.44 to 0.56; OpenEnv Arena, after filtering down layer by layer from about 90,000 findable RL environments, had only about 800 relativel
In the first quarter of 2026, approximately 128,000 micro-short dramas were released across the industry, of which AI micro-short dramas reached 122,000, accounting for more than 95%. In Hengdian, the number of short dramas that started production in the first quarter plummeted by three-quarters year-on-year, the studio vacancy rate approached 80%, and only 700 to 800 valid notices for ordinary extras were issued each day. The article compares the per-title cost of live-action short dramas, at 500,000 to 1 million yuan, with the assembly-line cost of AI short dramas, at 3,000 to 5,000 yuan, and argues that performing labor that can be broken down in a standardized way will be the first to be replaced.
In the first quarter of 2026, approximately 128,000 micro-short dramas were released across the industry, of which AI micro-short dramas reached 122,000, accounting for more than 95%. In Hengdian, the number of short dramas that started production in the first quarter plummeted by three-quarters year-on-year, the studio vacancy rate approached 80%, and only 700 to 800 valid notices for ordinary extras were issued each day. The article compares the per-title cost of live-action short dramas, at 5
In a Latent Space podcast interview, Anthropic engineering lead Katelyn Lesse and product lead Angela Jiang discussed the Opus 5.5 release, covering adjustments to the model's writing style, pricing based on intelligence per dollar, and API changes such as requiring thinking blocks to be retained across turns to prevent distillation. The interview also described Claude Managed Agents' approach of separating sessions and the harness from the sandbox that executes code, with credentials mounted separately through a proxy so that a sandbox failure does not take down the entire session; Dreaming, meanwhile, is implemented as one managed agent inspecting another agent's history and writing back to memory. The team said that as safety work for model releases has grown heavier, recent efforts have shifted more toward safety, with progress on features such as skills and memory slowing relatively.
In a Latent Space podcast interview, Anthropic engineering lead Katelyn Lesse and product lead Angela Jiang discussed the Opus 5.5 release, covering adjustments to the model's writing style, pricing based on intelligence per dollar, and API changes such as requiring thinking blocks to be retained across turns to prevent distillation. The interview also described Claude Managed Agents' approach of separating sessions and the harness from the sandbox that executes code, with credentials mounted se
a16z partners Ben Horowitz and Michael McGuiness wrote the article, using Databricks' four generations of technological evolution—from Spark and Lakehouse to Lakebase and agents—as a thread to discuss how agents are changing enterprises. The article states that Databricks has surpassed $7 billion in annualized revenue, up 80% year over year, at a valuation of $190 billion, with more than 17 million databases launched by agents every day. The authors argue that enterprises need an agent-facing system of record that captures definitions, permissions, and evaluation results, and believe that in the future more people's work will shift toward specifying tasks, delegating execution, and checking results.
a16z partners Ben Horowitz and Michael McGuiness wrote the article, using Databricks' four generations of technological evolution—from Spark and Lakehouse to Lakebase and agents—as a thread to discuss how agents are changing enterprises. The article states that Databricks has surpassed $7 billion in annualized revenue, up 80% year over year, at a valuation of $190 billion, with more than 17 million databases launched by agents every day. The authors argue that enterprises need an agent-facing sy
Amazon CEO Andy Jassy said in an interview on the a16z podcast that the company's capital expenditure is expected to reach $220 billion in 2026, but compute capacity is already sold out for 2026, essentially fully booked for 2027, and a large share of 2028 quotas has been locked in. He described enterprise AI demand as dumbbell-shaped: a large number of production systems in the middle have yet to be deeply integrated because inference costs are too high, and costs need to fall by another order of magnitude before an explosion in adoption occurs. Jassy also noted that 99% of AWS's top 1,000 customers use Graviton, which offers about 40% better price-performance than x86, while Trainium has a 30% to 40% advantage over comparable hardware.
Amazon CEO Andy Jassy said in an interview on the a16z podcast that the company's capital expenditure is expected to reach $220 billion in 2026, but compute capacity is already sold out for 2026, essentially fully booked for 2027, and a large share of 2028 quotas has been locked in. He described enterprise AI demand as dumbbell-shaped: a large number of production systems in the middle have yet to be deeply integrated because inference costs are too high, and costs need to fall by another order
In an interview, Claude engineering lead Felix Rieseberg said he barely writes code anymore; his core approach is to hand crash data, business tools, and a computer to the model and let it decide what to fix itself, rather than having a human act as a relay repeating things item by item. He suggests opening multiple threads on the same output dedicated to proving it wrong, then opening another as a judge to filter which criticisms hold up; rather than giving each Agent a tiny slice of data, it is better to give a panoramic business view and well-documented tools. He also mentioned that the half-life of Skills is extremely short, that both official guidance and his own advice are to use them cautiously, and that self-evaluation within the same session remains unreliable, so external signals should be sought as much as possible.
In an interview, Claude engineering lead Felix Rieseberg said he barely writes code anymore; his core approach is to hand crash data, business tools, and a computer to the model and let it decide what to fix itself, rather than having a human act as a relay repeating things item by item. He suggests opening multiple threads on the same output dedicated to proving it wrong, then opening another as a judge to filter which criticisms hold up; rather than giving each Agent a tiny slice of data, it i