🧠 I/O 2026 ↗
Google dropped a hefty stack of AI updates: Gemini Omni, Gemini 3.5, Gemini Spark, agentic Search, Universal Cart, Ask YouTube, Google Pics and intelligent eyewear. It felt less like a keynote and more like a rolling buffet with too many wheels.
The core pitch is clear - Google wants Gemini to stop being “a chatbot you visit” and become something quietly handling jobs across its products. Handy, a little eerie, probably inevitable. (blog.google)
🔎 A new era for AI Search ↗
Google says Search is getting its biggest box upgrade in over 25 years, with Gemini 3.5 Flash becoming the default model in AI Mode globally. The new Search box can take text, images, files, videos and Chrome tabs as inputs - proper multi-input mayhem, in a good way.
The more interesting bit is agents inside Search. Information agents will watch the web for updates, while agentic booking and mini-app style dashboards are coming too. Search is turning into a tiny operating system, or so it seems. (blog.google)
🧑🔬 OpenAI co-founder and former Tesla AI executive Karpathy joins Anthropic ↗
Andrej Karpathy has joined Anthropic, landing on the pretraining team behind Claude’s core model work. That is a serious talent win - not flashy-product serious, but “move the model frontier” serious.
He said he wants to get back to R&D, while keeping his education work on the back burner for later. The timing is neatly telling too, because frontier labs are now fighting over researchers like football clubs fight over strikers. (Reuters)
🏭 Mistral AI buys Austrian physics AI startup in industrial push ↗
Mistral AI acquired Linz-based Emmi AI to strengthen its industrial AI stack. Emmi’s thing is physics-heavy modelling - airflow, heat transfer, material stress - the stuff that matters when AI leaves the browser and bumps into physical machines.
Mistral is pitching this as a European manufacturing advantage, especially for aerospace, automotive and semiconductors. Not as glamorous as a talking assistant, maybe, but factories are where the money quietly wears steel-toe boots. (Reuters)
🧩 Google DeepMind hires staff from Contextual AI in licensing deal, source says ↗
Google DeepMind agreed to hire more than 20 researchers from Contextual AI and license its technology, according to a person familiar with the deal. The reported price tag sits around $80 million to $90 million.
This is another “not quite an acquisition” AI deal, which is becoming a whole genre now. Big labs get talent and tech, startups get a lifeline, regulators get a headache-shaped parcel. (Reuters)
💻 OpenAI and Dell Technologies partner to bring Codex to hybrid and on-premises enterprise environments ↗
OpenAI and Dell are partnering to bring Codex closer to enterprise data inside hybrid and on-premises environments. OpenAI says more than 4 million developers now use Codex every week, which is, frankly, massive.
The move is aimed at companies that want agents near codebases, documentation, systems of record and operational knowledge without flinging everything into the cloud. Sensible, not sexy - but probably exactly what cautious enterprises asked for. (OpenAI)
🏦 StanChart to cut over 7,000 jobs, boost AI to replace 'lower-value human capital' ↗
Standard Chartered plans to eliminate more than 7,000 jobs as it pushes AI and automation through corporate functions. The bank says the cuts equal about 15% of those roles, with back-office centres among the most affected.
The phrase “lower-value human capital” landed with all the warmth of a wet stapler. Staff will be offered reskilling options, but this is still one of the clearer examples of AI being tied directly to white-collar job cuts. (Reuters)
⚖️ Musk's failed court attack on OpenAI could leave lasting scars on Altman's reputation ↗
OpenAI beat Elon Musk in federal court after a jury rejected his case on timing grounds, easing a major legal threat around a possible IPO. That is the win. The catch - the trial dragged a lot of internal governance clutter into public view.
The ruling may help OpenAI’s business path, but the reputational bruising around Sam Altman is not nothing. Victory, yes. Clean victory? Not quite, more like a trophy with fingerprints all over it. (Reuters)
FAQ
What were the biggest AI updates from Google I/O 2026?
Google’s I/O 2026 updates centred on embedding Gemini more deeply across its product ecosystem. The announcements included Gemini Omni, Gemini 3.5, Gemini Spark, agentic Search, Universal Cart, Ask YouTube, Google Pics and intelligent eyewear. The direction is clear: Google wants AI to feel less like a separate chatbot and more like a built-in assistant woven through everyday workflows.
How is Google changing AI Search with agents?
Google is moving Search toward a more agentic experience. AI Mode is becoming more central, with Gemini 3.5 Flash used as the default model globally. The Search box can now accept text, images, files, videos and Chrome tabs as inputs. Google also described information agents that monitor the web, along with booking agents and mini-app-style dashboards.
Why does Andrej Karpathy joining Anthropic matter?
Andrej Karpathy joining Anthropic is notable because he is a major AI researcher with experience at OpenAI and Tesla. He is joining the pretraining team behind Claude’s core model work, which points to a focus on foundational research rather than a showy product role. It also reflects how competitive hiring has become among frontier AI labs.
What does Mistral AI’s Emmi AI acquisition mean for industrial AI?
Mistral AI’s acquisition of Emmi AI points to a stronger push into industrial AI. Emmi focuses on physics-heavy modelling, including airflow, heat transfer and material stress. That kind of work matters for sectors such as aerospace, automotive and semiconductors, where AI has to support physical-world engineering decisions, not just digital tasks.
Why are companies using licensing deals instead of full AI acquisitions?
In many cases, big AI companies want access to talent and technology without completing a traditional acquisition. Google DeepMind’s deal involving Contextual AI staff and technology is one example of this pattern. These arrangements can help large labs move quickly, while giving startups a commercial path. They can also draw scrutiny from regulators.
How is AI affecting enterprise jobs and workplace automation?
AI is increasingly being linked to direct workforce changes, especially in back-office and corporate functions. Standard Chartered’s plan to cut more than 7,000 jobs while expanding AI and automation is a clear example. The bank mentioned reskilling options, but the story still shows how AI adoption can affect white-collar roles, not just technical workflows.