What changed in AI today

• Mod authors must declare whether AI was used in their work when uploading new mods.
• AI-assisted mods must include a public explanation of how and where AI was used.
• Existing mods have 60 days to comply, with enforcement starting January 2nd.
• Site will not ban AI-generated mods outright, and will not license mods for AI training.

Why it matters for marketers: Marketers and content teams working with PC game modding communities should check AI disclosure rules before publishing or partnering on Nexus Mods content.

• The Australian Financial Review first reported the cuts on Tuesday, according to the source.
• Canva told 1,271 people, or 22% of its staff, to change roles, relocate, or lose their jobs.
• It is the company's first round of job cuts since it was founded in 2013.
• The source headline ties the cuts to AI rollout problems, but the text provided does not confirm that link.

Why it matters for marketers: Content teams that rely on Canva should watch for possible changes to product development, support, or service continuity after a 22% staff reduction.

• Shares of Synopsys and Cadence Design Systems dropped as much as 4% on Tuesday.
• The Information reported Anthropic told investors it uses its Claude models for engineering and design work on custom chips.
• The report says Anthropic's engineers found Claude reduces design time, according to the source text's partial detail.

Why it matters for marketers: Content teams covering chip design software should note that AI model makers moving into in-house chip engineering may be shifting how EDA vendors are valued.

• PIMCO CEO Emmanuel Roman said an AI market downturn is coming, per a Bloomberg roundup from a Hong Kong investment summit.
• Roman estimated the AI buildout could need $2 trillion in financing over five years; the US bond market is about $12 trillion.
• Meta reportedly drew $58 billion in orders for a $15 to $20 billion bond offering, and Microsoft sold $25 billion in bonds.
• Roman named US rate hikes, war, or a eurozone crisis as possible triggers for a market reversal.

Why it matters for marketers: Heavy AI debt and spending may face market risk, so content teams should expect possible budget pressure from AI-focused clients and vendors.

• Converted jet engines, rebuilt to burn natural gas, let some AI data centers make their own power while grid waits stretch for years.
• Rystad Energy's Audun Martinsen says a jet-powered unit can have a data center running in 24 months.
• Connecting 100 megawatts to the grid in Northern Virginia takes about seven years, according to the International Energy Agency.
• The engines are less efficient and pollute more than conventional gas turbines, so experts see them as a possible short-term option.

Why it matters for marketers: Content teams covering AI infrastructure should expect faster power-timeline stories and scrutiny of efficiency and emissions claims tied to these builds.

• AMD, an AI server chip designer, made the forecast in a statement on Tuesday.
• The projected growth is 20 percentage points faster than its third-quarter revenue growth.
• AMD also projected its gross margin would tick up 1 percentage point; the excerpt omits the timing.

Why it matters for marketers: Marketers covering AI chip demand can cite a specific AMD growth forecast, but should verify the margin timing before using it.

• FlashProve uses cryptographic proofs to show an output came from a specific model, even if the lab running it is the impostor.
• Labs would publish a model fingerprint that avoids revealing weights, so commercial closed models could participate.
• Verification could run on a local device, and code and a technical specification are on GitHub under an open-source license.
• Anthropic says the work is intermediate, with no plans to deploy it at Anthropic or any other lab yet.

Why it matters for marketers: Marketers citing AI-generated work or model performance may eventually be able to verify which model produced it, though deployment is not yet planned.

• Filings with the SEC show $360M for a third flagship fund, matching its 2024-raised predecessor.
• A separate $130M vehicle, Ballistic Ventures Fund B, is also being raised; neither fund had recorded sales at filing.
• Per JPMorgan Chase report, AI-enabled companies were about 72% of U.S. cybersecurity venture deals through May, up from 36% in 2019.

Why it matters for marketers: Marketers selling to security investors and startups can expect more cybersecurity VC capital and AI-focused deal activity as these funds close.

• Revenue grew 96% year over year to $2.31 billion, first time above $2 billion, beating the $2.18 billion expectation.
• U.S. commercial revenue ($984 million) surpassed U.S. government revenue ($953 million) for the first time.
• Total U.S. enterprise contract value fell to $1.46 billion from a record $2.13 billion in the prior quarter.
• CEO Alex Karp said enterprises sign faster but often on shorter contracts, with the ability to expand them.

Why it matters for marketers: Enterprise AI buyers are trying software quickly on short terms, so content teams selling AI platforms may need to prove ongoing value rather than rely on initial trials.

• Images circulating on X, Threads and Bluesky show both men with much more hair than usual.
• The Verge found watermarked originals via Google reverse image search; Bezos's dated 2023, Statham's 2011.
• No Content Credentials or SynthID markers were found, so whether AI tools were used is unknown.

Why it matters for marketers: Marketers using celebrity imagery should trace photos to original sources with reverse image search before reposting, since AI labels on social platforms often miss edits.

• Ukrainian drone strikes damaged Russian data centers, disrupting cloud services for millions of Russian users, per the report.
• Putin's November 2 decree requires operators to reserve 10 percent of space and capacity for possible government use, according to Meduza.
• A decree also lets a government commission authorize foreign data centers, a shift from the Sovereign Internet Law's restrictions.
• Yandex reportedly had to relocate all three of its main supercomputers after its AI data centers were targeted.

Why it matters for marketers: Marketers relying on Russian cloud platforms face service disruption and possible state capacity claims, so audience reach and campaign hosting there may be unreliable.

• Three optimizations applied together run MoE inference on a single DGX Spark machine.
• Ant Group inclusionAI's 124B Ling-3.0-flash model saw GSM8K decode throughput rise 42.5%.
• No detectable accuracy change was reported on the GSM8K benchmark.

Why it matters for marketers: Teams running large mixture-of-experts models on a single workstation-class system may get faster decoding without a measured accuracy cost.

• Soniox STT v6 is its newest speech-to-text API generation, aimed at voice agents.
• The model reads the agent's own spoken text while transcribing the caller, helping it recognize short replies.
• It uses that agent text to filter out the agent's voice, reducing echo in calls.
• Ships as stt-rt-v6 for streaming and stt-async-v6 for asynchronous use, per the source.

Why it matters for marketers: Teams building voice agents or call-based customer experiences may find the streaming model's echo handling relevant for evaluating speech input quality.

• Developers can integrate Siri AI features in all apps through the iOS 27.3 Developer Preview, rolling out Tuesday.
• New "app skills" let developers combine several tasks into one action for Siri AI to run.
• Users reportedly won't need to name an app while it is in the foreground, in a Live Activity, or in a Siri AI conversation about it.
• Apple's own apps, including Mail, Maps, Photos and Notes, gain the same expanded Siri AI capabilities.

Why it matters for marketers: Content and app teams can plan for voice-driven, multi-step actions inside their apps, so in-app flows may need rethinking for Siri AI requests.

• OpenAI's hardware team reportedly used GPT-6 Astra to design a device for talking to the Codex coding agent.
• Astra handled enclosure, circuit board layout, firmware and user interface; a contract manufacturer then built prototypes in under a month.
• Several revision rounds were needed because the model missed manufacturing constraints, tolerances and assembly details.
• On the BenchCAD 2.0 benchmark, an unnamed internal model scores 0.875 versus 0.594 for Astra at its highest reasoning level.

Why it matters for marketers: Teams building hardware or CAD workflows can expect AI to draft full designs quickly, but still need human review for manufacturability before prototyping.

• On its Q3 2026 earnings call, Google said it is building one seamless experience across AI Overviews, AI Mode and traditional Search.
• Search advertising rose 12% year over year, below the 15% Wall Street expected, while Alphabet revenue grew 25%.
• Google announced the Connected Apps Platform on the call day, letting Gemini use apps through MCP, including from within Search.
• The Gemini app reached one billion users, which Google described as the fastest growing product in its history.

Why it matters for marketers: Content teams should plan for agentic tasks inside Search, where website functions may run without a visit, while ad formats in AI Mode are still expanding.

• Owen Queen and James Zou tested 50 claims against 20 models released between July 2024 and September 2026.
• Among 22 findings with an estimated expiration date, the average lifespan was just 15.7 months.
• Claims about model limitations expire fastest, with five of nine reasoning-related findings considered expired.
• Findings about preferences and tendencies, such as sycophancy and self-preference, have lasted longer.

Why it matters for marketers: Date and name the model behind any LLM claim you publish, since 39 percent of tested findings had expired within two years.

• Bessemer Venture Partners led the Series F round, with Nvidia, Sound Ventures, and T. Rowe Price participating.
• Series E and F rounds together raised $1.65B; Sequoia led the earlier round.
• Fal reportedly surpassed $800M in annualized revenue earlier this year, per Bloomberg.
• Rival inference providers Baseten, Fireworks AI, and Modal also raised large rounds, but their gross margins remain thin.

Why it matters for marketers: Inference platforms are drawing heavy investment, so content and marketing teams relying on hosted AI video, image, and audio models may see more vendor consolidation and pricing pressure.

• LinkedIn says it has flattened the cost of running its AI-powered Hiring Assistant as usage grows.
• The Hiring Assistant uses large language models to automate candidate search.
• The source describes the cost pressure: more AI feature use means more server spending, which can squeeze margins.

Why it matters for marketers: Content and product teams running AI features may gain a cost-control approach to weigh when choosing which models to serve.

• Adam Fry, OpenAI's consumer product lead, made the claim at a Monday press event. He did not specify the source of Google's figure.
• OpenAI counts a search as any query that triggers a web search, excluding definitions and simple how-to explanations.
• OpenAI said last month it would show images of sponsored products in ChatGPT ads, not just a name, logo and link.
• Fry said ads have not hurt product usage, and OpenAI reportedly eyes an IPO potentially slated for early next year.

Why it matters for marketers: Marketers should expect ChatGPT to become a bigger search channel, with sponsored product visuals now coming to its ads.

• According to Binance, an engineer ran malicious code on a work computer while completing a task for a fake job interview.
• Binance says the attackers took several steps starting in June before gaining access to its system.
• The source says the hackers then used an AI tool, though the excerpt available does not give details.

Why it matters for marketers: Content and hiring teams should treat candidate assignments and unsolicited job tasks as a possible security risk, and check them before running any code.

• Studio Daimon's Food Devils was planned for early access today but still awaits Steam approval, after a build submitted October 12.
• Publisher 2 Left Thumbs said its Wheel Make It launch slipped by nearly two weeks in October due to build review delays.
• Valve says store page reviews typically take 3-5 business days and advises submitting at least 7 business days before launch.
• Developers suspect AI-generated 'slop' is clogging Steam's review queue; Valve had not replied to a request for comment.

Why it matters for marketers: Content teams planning game marketing around launch dates should build in weeks of buffer, since Steam review waits may exceed Valve's stated timelines.

• Experimental AI Discover tab sits in the TV app's top menu and surfaces shows by theme, such as "Quick Watches."
• Users can flag recommendations they dislike, and Netflix says the app will remember that feedback.
• Rollout is gradual, starting with US members who use Netflix in English on supported TVs.
• The launch follows reported growth slowdown, around 1,000 layoffs, and reportedly lower viewing time.

Why it matters for marketers: Marketers promoting Netflix titles may need to target theme-based discovery placements, as recommendations now cluster around moods and occasions rather than genres alone.

• Benchmark has four difficulty levels, from fixed values to swap-based reassignments, and tests the same problem in several formats.
• Frontier models including Gemini 3.8 Flash and GPT-6 Sol performed well on simple levels but fell sharply on harder ones.
• On the longest general finite problems, frontier model accuracy dropped to 13% or lower as inputs grew.
• Small Gated DeltaNet and hybrid models trained directly on StateBench stayed near perfect far beyond training lengths, while transformers never learned the hardest level.

Why it matters for marketers: Content teams using long-document AI should test models on the state-tracking level their workflows need, since frontier models still degrade on harder cases.

• Revenue reached $17 billion in the quarter ending September 30, up from $11.5 billion the prior quarter.
• Compute consumed 58% of revenue, up from 48% the quarter before, per an updated S-1 filing.
• Compute bill hit $10 billion, including $5.0 billion under short-term commitments that cost more than multi-year deals.
• Anthropic expects compute and infrastructure to remain its largest operating expense for the foreseeable future.

Why it matters for marketers: Marketers should expect AI vendor pricing and capacity to stay volatile, since compute costs are tied to infrastructure that could slow with credit or local pushback.

• Researchers still type in their commands, while AI agents steer the telescope and analyze incoming radio signals.
• Nvidia's Holoscan platform, used by SETI Institute and Breakthrough Listen, processes telescope data in real time.
• Nvidia says the hardware handles up to 2.4 terabits per second; the team says weeks of work now finish in hours.
• The team reportedly detected a Crab Nebula pulsar in real time with AI, and plans edge-capable open-source tools for future Moon sensors.

Why it matters for marketers: It shows AI agents running real-time signal analysis on live scientific data, a concrete case for teams weighing agentic systems in high-volume data pipelines.

• Palette uses LoRA on a 20B open-weight model, with one adapter split into a designer role for the brief and a coder role for slides.
• Training used about 1,000 synthetic decks, roughly 10,000 slides, in the IBM Design Language, checked by the same harness used at inference.
• On PresentBench, Palette ranks first on Economics and Education and second on the three-domain mean, 1.1 points behind NotebookLM.
• The harness adds 3.5 points on average across 131 decks, and parallel slide coding reportedly runs six to seven times faster.

Why it matters for marketers: Teams wanting template-matched decks can self-host an open model trained on their design language, keeping confidential content in-house.

• Starting next year, Google will turn off AI tools for K-12 students unless school administrators explicitly enable them.
• The change covers Gemini, NotebookLM, Gemini in Classroom, and Google Search, per TechCrunch's report.
• Schools that restrict AI entirely will also lose access to Google Search, since its AI features are now integrated.
• Google says it will share more specifics with schools in the coming months before changes roll out.

Why it matters for marketers: Content and edtech teams serving schools must plan for opt-in AI settings and check how Search access shifts if districts restrict AI.

• Simon Willison ran Qwen 3.8 27B locally in LM Studio on a laptop, testing the paper's no-tools weight-reading task.
• On the Pac-Man weights, the model took 46m 12s and 43,506 reasoning tokens, guessing an 8x8 maze game.
• It reportedly identified Pong after 1h 4m 54s, while it was still working on the Freeway prompt at writing time.
• The model is about 17GB and Willison says the paper's approach needs no tools, only reasoning over numbers.

Why it matters for marketers: Content teams testing local models should expect long reasoning times and context settings to be tuned before results are usable.

• The Information reports many in Silicon Valley are afraid of Nvidia CEO Jensen Huang.
• Nvidia has substantial money and influence, and Huang reportedly has the ear of several top government officials, including President Trump.
• The piece says the concern matters more for model developers like OpenAI, but the excerpt provided does not include the specifics.

Why it matters for marketers: It signals that Nvidia's influence over AI developers and policymakers may shape the vendor landscape content teams track and write about.

• Waller, co-CEO of Whalar Group, argues mutual respect is a more useful standard than authenticity for creator partnerships.
• Creators are increasingly involved from the start of strategy, contributing ideas, audience insight and sometimes ownership stakes.
• Waller expects AI to grow personalized content volume, making trusted human tastemakers more valuable and generic content harder to compete with.

Why it matters for marketers: Marketers should vet creator deals on mutual respect and shared brand goals, and involve creators earlier in strategy rather than adding them after campaigns are written.

• Senior research scientist Noam Brown shared a news story on OpenAI's changing stance on advanced chip sales to China.
• Policy leaders replied by citing earlier company posts, including a June safety blueprint on frontier AI.
• The exchange is reported by The Information, which frames it within a broader piece on Nvidia's soft power.

Why it matters for marketers: Content teams covering AI policy should note that company positions on China chip sales may be shifting, and that public policy documents are being used to answer internal questions.

• Elon Musk has given at least $276 million to Republican groups since July, per The New York Times.
• Leading the Future, an AI-industry super PAC, has spent $45 million; Public First Action says it raised $80 million.
• Tech billionaires reportedly gave over $340 million to fight Proposition 40, a one-time 5 percent billionaire tax in California.
• Prop 45 would shorten environmental reviews for eligible projects, which critics say could help data center developers avoid scrutiny.

Why it matters for marketers: Political ad and messaging budgets are shifting toward data centers and AI safety, so brands in those sectors should expect heavier scrutiny and crowded airwaves.

• On 90 AppWorld tasks with the same open-weight model, task completion ranged from 30% to 81% by harness.
• Scenarios where all three variants passed: 20 for CUGA, 8 for Hermes, and 2 for Deep Agents.
• CUGA is Apache 2.0 and can run in air-gapped sovereign clusters, with policies enforced by the runtime, not prompts.
• The authors caution the 90-task sample may not generalize, and CUGA used an AppWorld configuration the others did not.

Why it matters for marketers: Enterprise agent projects should weigh harness choice alongside the model, and put payment and data rules where the runtime enforces them, not in prompts.

• Apple will require all apps with dating or romantic relationship features to carry an 18+ rating, per a developer announcement.
• Starting November 16, developers must include new content descriptors so age ratings reflect app features.
• Apps must also disclose if they share a user's real-world location with another person or group.
• Brazil lets family location-sharing apps request an official rating from the Ministry of Justice instead of an A16 rating.

Why it matters for marketers: Marketers running dating, companion, or location-sharing apps should review age ratings and content descriptors before the November 16 deadline.

• Pickles, co-founder and former CTO, left the company and its board in 2023 and has advised since then.
• The council will work with engineering and product teams on the core technology and product roadmap.
• The announcement comes a week before The Trade Desk's expected Nov. 9 third-quarter earnings report.
• The Trade Desk did not say whether Pickles returns full time, rejoins the board, or holds an officer title.

Why it matters for marketers: Marketers relying on The Trade Desk's platform should watch whether its technology roadmap, including AI agent tools, shifts under renewed engineering leadership.

• AI token bills have skyrocketed, pressuring CFOs and boards to show spending boosts revenue or profits.
• Dan Diasio, EY Global AI Leader and Consulting CTO, has advised the world's largest companies on AI transformations.
• Jessica Lessin, The Information's Editor-in-Chief, interviewed Diasio in a live session titled AI Agenda Live.

Why it matters for marketers: Content and marketing teams selling AI tools will face buyers demanding measurable revenue or profit impact, not just capability claims.

• Writer says the Tuesday and Thursday format had become habit, so the format, content and posting schedule will vary until year end.
• Goal is to post what interests the writer, after recent conversations with Stripe, Granola, Zed and Raycast about building in AI.
• Writer says they now use AI as a tool for fun rather than to optimise everything, and read more long-form content.

Why it matters for marketers: Content teams may see this as a signal that a long-running newsletter is loosening its fixed schedule and format while the writer experiments.