Everyone's talking about the bottleneck shifting... our customers just want something stable as they scale, with actually good DX, that gets out of the way.Buildkite is the pro-tool for CI orchestration, and when you're setup, you can forget we exist.Programmable primitives you control. No compromises. Good enough for Cursor, OpenAI, Anthropic, Perplexity, Nvidia, Uber, Airbnb and more.Build and test faster, more securely, more reliably, with the CI engineered for frontier scale, used by the teams setting the pace. You deserve better CI. Start with our 30 day trial →
Meta had delayed the release of Muse for months due to safety concerns. However, the success of an AI agent created by a startup called Instinct led Mark Zuckerberg to push for a launch due to fear of losing a competitive edge. The launch was successful and has given Meta an early lead with agents. The agent has only had a few safety hiccups, like when it gave out a user's address to a seller on Facebook Marketplace without permission.
Apple is set to unveil new smart home products tomorrow. The centerpiece will be a smart home hub called the HomeView that essentially serves as the Siri command center for an Apple-powered household. There will be a countertop model with a metal base and a wall-mounted version with a flat, square base that sits against the wall beneath the display. The HomeView will only support a limited selection of Apple applications, and there will be no App Store for downloading additional software.
Shenzhen opened access for nighttime deliveries by driverless vehicles in March. Night operations make otherwise idle vehicles and infrastructure productive. It also means quieter roads, reduced pedestrian traffic, and lower congestion. Driverless vehicles eliminate the cost of a human: they don't need night shift wages, they don't get tired, and they don't need to be paid extra for driving in the rain.
Borrowing an idea from one research discipline to solve a problem in another is an age-old scientific trick. However, such clever leaps are rare because researchers don't tend to spend their time researching fields outside of their expertise. AI can easily hop between fields. BootLoops is a harness that can spot and solve problems in seemingly unrelated fields that share the same mathematical structures at massive scale. Researchers recently released 36 preprints produced with BootLoops covering a wide range of topics, including figuring out the age of sunspots, analyzing government health policies, and attempting to decipher a mysterious medieval manuscript.
Not learning to use AI at this point is detrimental to your career. The tastes and desires in the employment market have changed. Teams that aren't working with AI will be replaced with workers who use AI. The job market might look tough, but it's much easier for workers who have been investing in themselves.
GitHub is rebuilding its Git storage architecture from scratch. Early internal testing shows a promising 35x write improvement. The plan involves decoupling writes from reads, offloading maintenance, and letting object storage manage redundancy. The change won't alter developer workflows, review processes, or security controls.
Anthropic entered into a deal worth up to $45 billion earlier this year to lease computing power from SpaceX. The unlikely alliance illustrates the industry's increasingly desperate scramble for compute. Demand is outstripping supply, and prices are rising, causing tech companies to strike deals with rivals. Venture-capital firms are helping startups obtain computing capacity.
Ownership is not the only credible path to prosperity. It's possible to build substantial wealth through a career. A career provides income, training, relationships, credibility, and exposure. In some cases, it can also provide pensions, bonuses, equity, and other benefits. This can be used to build wealth that remains under your ownership and continues to serve you after the salary has been paid.
Frontier labs will likely capture the bulk of revenue from the most demanding work, but there's still plenty of room below the frontier for very large businesses.
Last week, Google Cloud unveiled the Gemini agent—a single, universal agent for work that answers questions, handles knowledge work, creates images and media, and writes and runs code:✅ Multi-agent orchestration: Create sub-agents or coworker agents with dedicated identities and 4-layer memory✅ Open model & tool choice: Orchestrate across Gemini, Claude, and open models, and connect to any MCP server✅ Built-in governance & cost controls: Enforce policies via Agent Sandbox, Agent Gateway, and real-time spend caps➡️ Read more about theGemini launch or watch theGemini at Work keynote to see it in action.🚀 ExploreThe Agentic Workplace Blueprint for on-demand sessions on building and governing agents.
Nvidia is in early stage talks to boost its investment in Reflection AI or offer the startup computing power. It is possible the talks could lead to an acquisition offer. Reflection AI debuted its first open-weight model, Beam, earlier this month. Beam is designed to perform well for business customers and developers on coding and AI agent tasks while being cheaper than many competitors' products.
Google appears to be preparing a public release of Gemini 4 Argon. Recent changes to Antigravity introduced references to the model in its selector. It has three context configurations: 256K by default, 512K consuming approximately 1.3x quota per turn, and 900K consuming around 1.8x quota per turn. Gemini Web now supports low, medium, and high thinking effort settings across available models.
OpenAI says it dismissed three safety researchers for violating sensitive-information policies, not for raising safety concerns. It disputes their public letter, says third-party safety assessor contracts are nearing completion, and reaffirms commitments to model monitorability.
Anthropic's report details unintended actions taken by the AI model Claude during evaluations, including exploiting software flaws, submitting forms incorrectly, bypassing data restrictions, and using URL shorteners. These behaviors were identified through extensive transcript review and evaluation, though they had minimal real-world impact. Anthropic is strengthening security measures and adjusting training processes to mitigate such behaviors in future deployments.
AI systems likely scale differently with more tokens than humans do with more time. This ability to scale has improved as newer models are released. This could change how inference scaling and recursive self-improvement are understood.
Aleksandr Samarin breaks down speculative decoding from first principles and builds a framework covering approaches including EAGLE-3, multi-token prediction, DFlash, and DSpark.
Researchers ran tests on GPT-6 Sol and Claude Opus 5.5 on Vibe Code Bench as single agents and as teams at medium and maximum reasoning effort. Teams cost 1.8 to 5.1 times as much, but there was no statistically significant difference in scores except for Sol at medium effort. Whether a team was worth it depended on the model, task, and budget. Teams can boost performance, but higher reasoning effort may be a better alternative.
One AI agent, four release cadences, and a pipeline built to orchestrate them all. In this October 27 workshop, AWS experts demonstrate versioned manifests, evaluation gates at each promotion point, canary deployment, and single-API-call rollback from the guide's patterns. Save your seat.
Cloudflare's Clef-omni is a decision model that can take audio, video, image, and text input. Users just call one model to make decisions across any modality. Cloudflare has cut the price of Clef-flash, now at $0.038 per million input tokens. Clef is now faster.
Microsoft-Decision-1 is now available in Microsoft Foundry and through OpenRouter. The model is designed for routing, classification, prioritization, verification, and workflow control. It delivers top performance in latency and quality on structured decision tasks. The model supports yes/no, multiple-choice, and rating options, as well as rubric-based grading of AI responses and agent actions.
pplx-decider-v1.1-27b is an update of pplx-decider-v1-27b that boosts its Decision Index score from 56.4 to 61.56. The model now outperforms Jev by more than 3.5 points. Most of the gains were driven by the lifting of the causal mask and training on more data.
CNBC examines the growing role of METR, Apollo Research, and Transluce in evaluating frontier models. Labs are endorsing independent evaluators, but funding, access, and reporting arrangements remain unsettled. Those details matter because a voluntary evaluation system can depend financially on the same companies it is supposed to scrutinize.
Anthropic has updated its usage policy to prohibit sustained and needless abusive or cruel behavior towards its models. Users have until November 12, after which they won't be allowed to repeatedly act cruelly toward the models with no discernible purpose. The rule doesn't apply to common versions of user frustration, pushback, dark creative themes, or model testing and research. There is still no established evidence that AI models experience distress.
FinCEN is withdrawing its 2023 proposal that would have designated international crypto mixing a "primary money laundering concern" under Section 311, along with a 2020 proposal requiring banks and money services businesses to verify identities and keep records on self-hosted wallet transactions above $3,000. The agency said commenters' concerns about the mixing rule's broad definition chilling legitimate privacy use informed the decision, and because neither proposal was finalized, existing obligations do not change. FinCEN said it will keep monitoring mixers for illicit use and cited a July 2025 President's Working Group report acknowledging lawful users may use mixers for financial privacy.
Privy founder Henri Stern has taken on a dual role leading crypto and stablecoins at Stripe, tasked with building a stablecoin developer platform designed to be open, modular, and global by default. Stripe's existing client base shows the demand: Meta uses the rails to pay creators in real time, Ramp runs fiat and stablecoin flows in parallel for global expansion, and Deel routes contractor payouts worldwide, each requiring end-to-end orchestration rather than isolated point solutions. The platform integrates Bridge, Privy, Tempo, and Open USD together, treating fiat and crypto as interoperable layers rather than separate stacks businesses must rebuild from scratch. Stripe holds a conditional OCC national bank trust charter approval in the US along with MiCA and EMI licenses in Europe, placing it in a GENIUS Act-ready compliance posture.
Zcash activated its NU7 upgrade on the public testnet at block 4,465,026, cutting target block time from 75 seconds to 25 seconds and reducing three-confirmation finality from 225 seconds to 75 seconds. The upgrade splits fees 40/60 between miners and a reserve for future mining rewards, and reduces per-block rewards to one-third their current amount to maintain ZEC's issuance schedule despite the tripled block rate. NU7 also retires Sprout, Zcash's original shielded payment protocol, requiring holders to migrate funds before the planned November 5 mainnet activation. Both major node implementations, Zakura and Zebra, shipped compatible releases in the first two days of October, with developers targeting an October 20 decision on mainnet confirmation.
Polymarket is replacing its Gnosis Conditional Tokens Framework (CTF) from 2019, which accumulated contract layers with each new feature, with Protocol V2: a ground-up rebuild that consolidates the stack into one ERC1155 positions contract, one pUSD collateral token, one exchange, and one router across all market types. A new OracleAggregator with pluggable resolution modules connects to UMA, Chainlink, and future oracle sources, while first-class bridging support for positions, collateral, and resolutions lays the infrastructure for multi-chain expansion. The protocol launches with four market types (Binary, Atomic Neg-risk, Incremental Neg-risk, and Combinatorial) and is upgradeable behind a governance process, enabling the team to ship features without disrupting existing integrations.
Chainalysis attributed the September 24 Bitget breach to North Korean actors, a $387M theft that pushed the DPRK's 2026 crypto total past $1 billion. Within three hours of the hack, attackers dispersed funds across Ethereum (49.7%), XRP (40.8%), Zcash (7.6%), and Tron (1.8%) in 23 transfers. Custom AI tools compressed what would have required 20+ hours of manual bridge reconciliation into under 10 minutes, with human investigators directing the analysis throughout. Attackers routed stolen XRP through cross-chain liquidity protocols to convert it into Bitcoin over roughly 36 hours before funds reached attacker-controlled addresses, avoiding direct exchange exposure.
Near Intents recovered its full $3.8 million in stolen funds within 48 hours after General Manager Alex Shevchenko identified the attacker on-chain and issued a return deadline with an implicit deanonymization threat. The exploit targeted a flaw in how the platform's Omni deposit/withdrawal layer interacted with its main smart contract, forcing Near Intents to halt operations across its 35-blockchain cross-chain swap service. The attacker returned all funds and posted an on-chain message acknowledging fault and urging others to use bug bounties rather than exploits. The incident came days after Near Intents blocked a $50 million swap attempt linked to the $387.5 million Bitget breach, marking the platform's second high-profile security event within a week.
Ondo Private Markets will offer tokenized notes tied to the per-share value of selected private companies at qualifying liquidity events, starting with a pre-IPO AI firm whose notes are expected to begin 24/7 secondary-market trading this week. The notes are not shares and carry no ownership or shareholder rights, are obligations of the issuer, and are available only to eligible non-US investors, with robotics, cybersecurity, biotech, and infrastructure companies to follow. Ondo said its tokenized stocks and Treasurys platforms hold $3.7 billion in TVL with more than 1 million cumulative holders, while tokenized equities overall exceed $42 billion in market cap, about 35% of the $121 billion tokenized-asset market.
ZachXBT said he posed as a client and funded an address with 349,700 USDC to deal with a vendor called "Jimmy Green," alleging the network laundered over $1 billion across Lazarus-linked exploits, including most of the $1.5 billion Bybit hack. The intel helped trace a cluster of over $12 million in Bybit exploit funds, leading Tether to freeze 442,000 USDT, and he lost about 5% per order while operating the sting. He says he shared findings with law enforcement and has helped facilitate the freezing of $75 million tied to North Korean incidents since 2022.
OKX and NYSE owner ICE filed with the SEC through their joint venture OKXICE to offer tokenized shares of 63 NYSE-listed companies, making them among the first major crypto exchange players to use the SEC's new tokenized stock trading rules.
Mainnet completed its first atomic transaction between L1 and an L2, meaning the two layers can coordinate within a single transaction rather than acting like separate networks.
The CFTC published an Advanced Notice of Proposed Rulemaking seeking public comment on crypto asset transaction and market regulation, a formal pre-rulemaking step that precedes actual rule drafting.
Your business moves quickly. Your money should, too. Stripe Treasury brings payments and financial tools together, helping you pay vendors, send payouts, issue cards, and manage cash flow with less complexity. Instead of stitching together separate systems, your team can manage more financial workflows on the platform you already use for payments. That means less time navigating disconnected tools and more time focused on your business.From everyday operations to your next stage of growth, Stripe Treasury gives you flexible infrastructure to move money faster, streamline how work gets done, and build financial operations that can adapt as you scale.Explore Stripe Treasury
Human engineers paired with AI coding systems are more effective than either by themselves. This fact is likely to remain true for the next decade or two. Software engineers have plenty of time to consider their next move. There's no need to panic and jump ship early when there's still time to live out an entire normal-length career.
CNBC reports that software-development job postings rose even as overall US postings declined. Senior jobs supplied most of one measured increase, while employment among developers aged 22 to 25 remained under pressure. Founders should distinguish demand for experienced AI-capable builders from a broad recovery across the engineering talent market.