- Users can now generate AI content in Gemini and refine it in professional editors without switching apps.
- Canva is live for Gemini AI Ultra subscribers;
- Adobe and CapCut roll out in the coming weeks.
- Distribution is the moat — and Google is exploiting it.
Snapshot — May 25, 2026
48 stories
Zhe Zhu's doctoral dissertation argues that GenAI's biggest workforce risk is adoption lag, not displacement, and proposes an eight-step framework for moving organizations from experimentation to "AI-native" operations. Employees who view tools like ChatGPT and Gemini as collaborators are measurably more engaged than those treating them as threats — a structured counter-narrative useful for HR and change-management teams.
- DeepMind's AlphaProof Nexus, pairing Gemini 3.1 Pro with the Lean proof assistant, autonomously resolved 9 of 353 open Erdős problems and 44 of 492 OEIS conjectures, plus a 15-year-old algebraic geometry question.
- Each solved problem reportedly cost only "a few hundred dollars" in compute.
- The hallucination-control architecture — Lean's compiler verifies every step — offers a template for high-stakes reasoning systems where output correctness can be formally certified rather than benchmark-approximated.
- Anthropic is closing a $30B round at a $900B+ valuation as soon as the end of this week, co-led by Sequoia, Dragoneer, Greenoaks, and Altimeter.
- If the term sheet lands at the upper end, Anthropic surpasses OpenAI's $852B March valuation for the first time — a complete reversal from February 2026's $380B mark.
- Anthropic is in talks to adopt Microsoft's custom Maia 200 AI chip for Claude models, making Microsoft the fifth silicon partner alongside NVIDIA, AWS Trainium, Google TPUs, and SpaceX compute.
- Most labs lock into one chip vendor;
- Anthropic is treating compute optionality as a competitive moat.
Anthropic published Olah's Vatican remarks delivered alongside Pope Leo XIV at the launch of his first encyclical on safeguarding human dignity in the age of AI. Olah said AI development "operates inside a set of incentives and constraints that can sometimes conflict with doing the right thing," and urged religious, civil, scholarly, and government communities to push the field "in a better direction." A striking public alignment of the most safety-branded frontier lab with the Vatican.
The Apple–Google partnership announced January 12, 2026 — granting Apple access to a custom 1.2 trillion-parameter Gemini model purpose-built for Siri and Apple Intelligence — continues to drive industry analysis ahead of WWDC 2026 (June 8). Estimated at ~$1B/year, the non-exclusive licensing deal is being characterized by analysts as "the most financially sound decision Apple could have made," with the rebuilt Siri expected to ship in iOS 27.
- A newly discovered genai.apple.com subdomain surfaced over the weekend, reinforcing expectations of a major generative-AI announcement at WWDC on June 8.
- Industry watchers anticipate a Siri rebuild, expanded Apple Intelligence features, and deeper on-device model integration across iPhone, iPad, and Mac.
- Chinese models — Kimi K2.6, DeepSeek V4, GLM-5.1, Qwen 3 — now account for 60% of all AI usage on OpenRouter, the most-used third-party AI model router.
- The clearest single signal that the open-weights tier is now Chinese-led.
- Meta's delayed Avocado model — the last credible US open-weights frontier candidate — has gone silent.
- ClickUp's mass layoff is being read by analysts as a leading indicator for how productivity-software vendors are restructuring around AI agents.
- The story extends the May narrative — Meta cut 8,000 jobs starting May 20 — that hyperscalers and SaaS firms are trading headcount for AI compute capacity.
- Academic Research N Research
- Google DeepMind’s AlphaProof Nexus reportedly solved nine open Erdős problems and proved dozens of additional conjectures.
- The result reinforces the thesis that frontier AI systems are becoming research instruments capable of producing verifiable mathematical progress, not merely assisting with literature review or code generation.
- Salesforce, Snowflake, and Asana earnings are being watched as a referendum on whether AI-native startups are taking share from incumbents or whether incumbents can repackage AI into durable growth.
- The Cowork newsletter framed this as an important signal for CIOs because buying decisions may shift from seat-based software to outcome-driven AI workflows.
- The EU AI Act becomes fully enforceable on August 2, 2026 — the first comprehensive binding AI regulation in any jurisdiction.
- Penalty structure: up to €35M or 7% of global annual turnover for prohibited practices; €15M or 3% for high-risk violations.
- GPAI obligations for models above 10²⁵ FLOPs of cumulative compute — covering all current frontier models — include adversarial testing, incident reporting, and energy disclosure.
TechCrunch's feature argues that even hyperscalers are improvising AI security controls in production — prompt injection, agent permissioning, and tool-call exfiltration are being addressed reactively rather than through mature frameworks. The piece resonates with a growing CISO-side concern as enterprise agent rollouts accelerate.
IBM unveiled Bob, an AI platform aimed at controlling cost overruns across the software development lifecycle. The product targets enterprise engineering leaders who have struggled to attribute and forecast spend across AI-assisted coding, CI/CD, and observability — a growing concern as agentic workflows multiply infrastructure draws.
Lucra Sports founder Dylan Robbins landed Cathie Wood and her ARK Invest Venture Fund as the lead in a $20M Series B — despite Lucra being a white-label interactive-gaming-as-loyalty-program company, not AI. A rare counter-trend data point: the entire VC market is chasing AI, but a non-AI raise this size still closed in May 2026.
A Mayo Clinic study describes an AI screening model that surfaced pancreatic cancer indicators in patient records up to three years before the disease was clinically diagnosed. The result sits among a growing body of academic work — increasingly cited at AI policy hearings — making the case that medical-AI early-detection benefits should weigh heavily against blanket regulatory caution.
Coverage this week continued to digest the up-to-$50B Meta–NVIDIA compute arrangement, with analysts framing it alongside the OpenAI Stargate and Anthropic compute commitments as evidence that hyperscaler and frontier-lab GPU buy-side concentration is now the dominant driver of NVIDIA's forward revenue. Combined 2026 AI capex across the Magnificent Seven is tracking past $700B.
Microsoft confirmed full multi-model availability inside Microsoft 365 Copilot — users can now pick Claude Opus 4.7, GPT-5.5, or Auto mode per task. Agent 365 hit general availability May 1 at $15 per user/month, a new Shadow AI pane lets IT block unsanctioned AI via Intune, and the Researcher Critique feature pairs a GPT-class model with a Claude-class model for a built-in expert review pass.
- The Windows 11 April 2026 update adds an official uninstall path for Copilot through Settings, plus a new Group Policy ("Remove Microsoft Copilot app") for enterprise administrators.
- Microsoft cited weak consumer monetization — only 3.3% of eligible users converted to paid Copilot — and persistent IT-admin pressure.
- A new wave of Nemotron-Labs diffusion language models claims to compress text-generation latency to near-keystroke speeds, applying diffusion techniques previously confined to image synthesis.
- If validated, the result reframes streaming-chat and live-translation economics — but also stresses content-safety pipelines that depend on iterative validation.
- NextEra Energy announced a $67 billion deal to acquire Dominion Energy — the largest US utility merger in history — with AI-driven power demand as the primary strategic rationale.
- AI data centers are projected to consume 15–25% of US electricity by 2030.
- NextEra, operator of the largest renewable portfolio in North America, is acquiring Dominion specifically to build generation and transmission for hyperscale AI workloads.
- Nvidia disclosed an additional $80 billion stock repurchase authorization following Q1 results that beat both Wall Street consensus and the company's own guidance.
- The buyback signals management's confidence in continued AI-cycle demand.
- Separately, Nvidia disclosed $43 billion in startup holdings on its balance sheet — an indicator of how deeply the chip leader is now intertwined with the AI ecosystem it supplies.
- MarkTechPost published a hands-on guide comparing FedAvg and FedProx federated-learning algorithms on Non-IID CIFAR-10 using NVIDIA FLARE.
- Federated learning interest is climbing in 2026 as enterprises seek to train on regulated data — particularly healthcare and finance — without centralizing it.
- Directly relevant to Microsoft's Azure Confidential Computing positioning.
- OpenAI is preparing a confidential S-1 with the SEC, advised by Goldman Sachs and Morgan Stanley, targeting a public listing as early as September 2026 at a valuation above $1 trillion.
- The company generates $25B in ARR and 900M weekly active users — but is operating at a loss while Anthropic hits profitability.
- OpenAI quietly launched a beta ChatGPT add-in for PowerPoint, letting free and paid users build, edit, and refine slides from a sidebar inside the app — directly competing with Microsoft 365 Copilot's native PowerPoint experience.
- The integration extends ChatGPT's footprint inside the Microsoft Office surface area at exactly the moment Redmond is restructuring its Anthropic and OpenAI partnerships.
An internal OpenAI reasoning model autonomously produced a counterexample to Paul Erdős's 1946 unit-distance conjecture — the first time a frontier AI has overturned a long-standing open problem in combinatorial geometry. The result is being cited as a milestone for AI-assisted mathematics and is expected to accelerate adoption of frontier reasoning models in formal research workflows.
- OSCAR is an attention-aware 2-bit KV-cache quantization system designed to make long-context inference dramatically cheaper.
- The release matters for any team serving models above ~200K tokens, where KV-cache memory has become the dominant inference cost driver.
- The open-source posture is also a strategic move to commoditize a layer where hyperscalers currently extract premium pricing.
- Pope Leo XIV’s first encyclical, Magnifica Humanitas, uses AI to critique the concentration of technical and economic power, calling for stronger oversight and broader participation from communities affected by the technology.
- TechCrunch’s analysis emphasizes that the document is not only about AI, but about the older governance failures AI could amplify.
Pope Leo XIV released "Magnifica Humanitas," an encyclical letter setting out the Catholic Church's framework for AI development, with explicit teaching on human dignity, labor displacement, autonomous weapons, and the moral responsibilities of AI developers and deployers. The document is being read alongside the EU AI Inventory and U.S. executive order context as part of a broader normative consolidation around AI accountability.
The first papal encyclical on artificial intelligence — signed May 15 on the 135th anniversary of Rerum Novarum — calls for "disarming" AI, rejects autonomous lethal weapons ("no algorithm can make war morally acceptable"), and condemns the "race for ever more powerful algorithms." Anthropic co-founder Christopher Olah presented alongside the Pope at the Vatican Synod Hall. Expect this document to be cited in upcoming policy debates worldwide.
- Pope Leo XIV’s Magnifica Humanitas frames AI as a question of human dignity, labor disruption, democratic oversight, and concentration of power.
- TechCrunch’s analysis argues the encyclical is less about AI as a tool and more about the older governance failures that AI could amplify.
- Business Insider separately tracked reactions from technology and policy leaders, while The Information highlighted Anthropic’s role in the Vatican launch.
Alibaba shipped Qwen 3.7 Max with new reasoning and tool-use modes, while xAI launched "Grok Build," a paid developer tier targeted at agent and coding workloads. Both releases reinforce that frontier model leadership has fragmented along workload lines — coding, agentic execution, multimodal, long-context — and that procurement teams should expect to evaluate three to five vendors per workload type going into H2 2026.
- Salesforce, Snowflake, and Asana all report Q1 earnings this week, providing fresh data on whether AI startups are taking share from incumbents.
- Workday CEO Aneel Bhusri set the template last week, declaring: "With AI, we are essentially a startup again." Expect each incumbent to credit AI for any momentum and brush off any softness — making the underlying numbers the real signal.
- SoftBank insiders are reportedly raising concerns about whether Son's $60B OpenAI commitment can be justified given rising debt loads, asset sales and growing enterprise pressure from Anthropic.
- The story lands days after OpenAI's confidential IPO filing targeting a $1T valuation, sharpening focus on whether public-market numbers will live up to private-market enthusiasm.
# TechCrunch analysis: "The Pope's AI encyclical isn't really about AI"
Rebecca Bellan reads the 200-page encyclical as a critique of power concentration, citing Leo's line that "when such power is concentrated in the hands of a few, it tends to become opaque and evade public oversight," and his call to end the AI arms race. TechCrunch ties the timing to President Trump's recent delay of an AI executive order on the urging of VC David Sacks.
- Bellan argues Magnifica Humanitas is less a technology policy document and more a worker-rights manifesto using AI as the framing device — echoing Pope Leo XIII's 1891 Rerum Novarum on the Industrial Revolution.
- For policymakers and enterprise leaders, the document positions labor displacement (not existential risk) as the primary AI-ethics axis through 2026-2027.
- President Trump abruptly canceled the signing of an AI executive order, telling reporters it risked undermining America's competitive edge.
- The order would have created a pre-release vetting process for advanced models — a direct response to security concerns triggered by Anthropic's Claude Mythos.
- Axios reported that Mark Zuckerberg, Elon Musk, and David Sacks called the president directly in the hours before the scheduled signing.
- The Information’s Twilio coverage frames AI as both a growth driver and a source of new business-model pressure.
- The company’s AI-enabled products are strengthening its customer proposition, but the economics of model usage, contract structure, and software-vendor cost exposure remain in flux.
- The broader executive takeaway is that AI adoption can raise revenue relevance while simultaneously tightening scrutiny on gross margin and renewal terms.
UC Davis researchers described a miniature silicon spectrometer that uses 16 tuned photodiodes and a neural network to reconstruct spectral information computationally. The approach replaces bulky optics with AI-based reconstruction, opening a path toward lower-cost hyperspectral sensing for diagnostics, food inspection, pollution monitoring, and embedded devices.
- University of Vaasa research suggests generative AI can increase employee engagement and adaptability when workers view it as a collaborator rather than a threat.
- The research also warns that over-trust and under-trust both create risk: one weakens judgment, while the other leaves productivity gains unused.
WorkOS opened a proposed standard for registering AI agents as OAuth clients with enterprise identity providers. The protocol addresses one of the most-cited blockers to enterprise agent rollout: scoping, auditing, and revoking agent access at the IdP layer rather than the application layer.
- WSJ profiles the senior executive driving Meta's accelerated AI agenda — covering Muse Spark development, the company's recent restructuring and layoffs, and the playbook for translating Llama and consumer-AI bets into ad revenue.
- The piece is essential reading for anyone tracking how the second-largest ad platform on the planet is repositioning around AI.
- WSJ reports that AI deployment is moving beyond Fortune 500 budgets to small and mid-sized businesses — a structural shift that mirrors OpenAI's pivot to smaller advertisers and the broader commoditization of inference.
- Separately, WSJ flags that chatbot long-term memory features are creating new privacy and exposure risks for both individuals and enterprises.
xAI's News page leads with "Introducing Grok Build — Now in early beta for all SuperGrok and X Premium Plus subscribers. Grok Build is a new coding agent that runs right from your terminal." The CLI runs on Grok 4.3 beta with a 2M-token context window and up to 8 parallel subagents — xAI's direct play against Claude Code, OpenAI Codex, and Cursor's Composer.
- xAI made Grok 4.3 the default model option inside the NVIDIA-backed OpenClaw agent platform, accessed via OAuth.
- The integration creates a credible third-pole agentic stack alongside Anthropic's Claude Code ecosystem and Google's Gemini-Antigravity surface — and gives developers a frictionless way to A/B agents across model providers.
- Xreal, Google's official smartglasses hardware partner for the Android XR platform, says it has cracked the wearable category's long-standing tradeoff between weight, optical quality, and battery life.
- The reveal complements Google I/O's Gemini-powered Samsung XR glasses announcement and signals that smartglasses will be the next major AI hardware battleground.