AI in Flow

Claire and Peter break down a packed day of AI news on AI in Flow: the White House reportedly shelves a plan for pre-release access to frontier models, leaving companies to build their own testing and incident-response guardrails; Google’s AI-heavy Search redesign draws backlash and forces marketers to rethink click-through strategy and channel mix; Big Tech engages Vatican officials ahead of an influential papal AI document; OpenAI’s high-profile safety hiring signals how seriously labs are treating self-improving systems; Apple’s rumored genai.apple.com hints at a broader on-device AI push; and Merlin Labs’ AI copilot lands a major US Air Force contract after demonstrated flight capabilities. Plus: DeepSeek makes a steep model price cut permanent, while Singapore and India highlight the growing infrastructure and sustainability constraints behind the AI boom.

Claire and Peter break down a packed day of AI headlines on AI in Flow: Nvidia tightens the screws on export-control compliance across the AI server supply chain, a proposed US AI safety executive order reportedly collapses, and Washington eyes billions for secure compute as competition for GPUs intensifies. They also cover OpenAI’s new safety push around recursive self-improvement, Google’s decision to keep AI Overviews as a default layer in Search—raising fresh pressure on publishers and brands—and signals of a major Siri upgrade ahead of WWDC. Plus: new job-impact forecasts, a deepfake misinformation case that underscores the need for rapid-response plans, and Bumble’s move toward an AI matchmaking assistant.

Claire and Peter break down today’s biggest stories on AI in Flow: a reported $9B classified U.S. push for advanced AI chips and secure data-centre capacity for intelligence agencies (plus faster compute access and continued NSA use of Anthropic models); Amp’s $1.3B raise to build a “shared AI grid” for startups and universities; and the people impact of automation after Standard Chartered’s CEO apologises for comments as the bank plans back-office cuts. They also look at Microsoft’s consumer Copilot monetisation challenges amid a key marketing departure, Google’s AI Overview oddities that highlight search volatility, India’s expanding Global Capability Centres as work shifts toward higher-value in-house roles, and rising stakes in synthetic media—from alleged AI-driven defamation in South Korea to viral political deepfakes in the U.S. The episode closes with a hopeful healthcare use case: an AI ultrasound tool helping detect gallbladder cancer, and a reminder that AI’s investment narrative is becoming more nuanced.

In this episode of AI in Flow, Claire and Peter break down SpaceX’s IPO filing and what its massive pivot toward AI infrastructure and data-centre spending could mean for cloud and telecom competition. They cover Google’s rapid reversal on Gemini limits for its Antigravity coding suite, the White House delay of AI cybersecurity rules, and why the Pentagon is testing Google and OpenAI models after friction with Anthropic’s usage restrictions. Plus: Spotify’s new licensed AI remix and cover deals with UMG, the US Health Department’s plan to use generative AI to scale fraud and compliance audits, fresh evidence showing consumer chatbots still miss key diagnoses, and a promising breakthrough in AI-designed miniproteins for drug discovery. The episode closes with a look at policy and infrastructure pressures—from California’s job-displacement planning to Utah’s backlash over the energy and water demands of mega data centres.

Claire and Peter break down a packed day in AI as the story moves from flashy demos to capital, governance, and real-world deployment. They unpack SpaceX’s IPO filing and how AI investment, Musk-linked related-party ties, and control structures could reshape risk for partners—and what OpenAI’s reported IPO preparations might mean for enterprise contracts and pricing. Then they look at Washington’s proposed voluntary pre-release reviews for frontier models, Nvidia’s blowout quarter alongside its effective exit from China’s AI chip market, and Singtel’s push into sovereign GPU-as-a-Service. The episode also covers Intuit’s major restructuring, a milestone EU certification for an AI-native EHR platform, retail wins from cleaner product data pipelines, and a new Nature paper on AI systems that can build scientific software—faster than humans, with big implications for verification and safety.

Claire and Peter break down a packed day of AI shifts reshaping distribution, privacy, security, and infrastructure. Google pushes Search further into an agent that can act on your behalf while rolling out faster Gemini experiences—and OpenAI adopts Google’s SynthID watermarking, signaling a growing push for content provenance. YouTube’s new Ask YouTube feature can jump viewers straight to the answer, raising big questions for creators and brands reliant on watch time. Meanwhile, Google DeepMind’s acquihire-style deal highlights the battle for talent. On messaging and privacy, WhatsApp tests “disappear after read” messages and Meta introduces Incognito Chat for Meta AI—useful for users, but thorny for compliance and record-keeping. OpenAI expands globally with a major investment in an applied AI lab in Singapore. In cybersecurity, Verizon’s DBIR shows vulnerability exploitation overtaking stolen credentials, with AI accelerating attackers and “Shadow AI” emerging as a key data-loss factor. The episode also covers China’s strengthening domestic AI stack with Alibaba’s new chip and model preview, synthetic media advances with Gemini Omni Flash, growing publisher concerns about AI Search cannibalizing traffic, and an unexpected hardware bottleneck: MLCC component supply in AI servers. It’s a clear reminder that in AI, operational adaptation—not just experimentation—decides who wins.

In today’s AI in Flow, Claire and Peter unpack a wave of signals that AI is no longer a side project for the biggest players. Meta is reportedly moving 7,000 staff into AI-focused teams alongside planned layoffs, accelerating a shift toward AI agents and consumer AI features across its platforms. Google and Blackstone announce a $25B TPU-backed cloud venture, highlighting that enterprise AI strategy is increasingly a multi-cloud—and multi-silicon—decision. The hosts also cover Google’s stance on AI-assisted coding accountability, Anthropic’s updated approach to sharing AI-discovered vulnerability findings, and why compute scarcity may persist into 2027 as agentic AI scales. Plus: Dell’s surge in AI factory deployments and financing, evolving AI-native architecture needs, Nvidia’s China outlook, and Tencent Cloud’s push for low-latency multimodal agents in Asia.

Claire and Peter break down why AI is getting more capable—and more operationally risky. They cover new research showing how easily major model guardrails can be bypassed, what the Musk vs OpenAI trial reveals about control and governance in frontier AI, and Apple’s reported Siri reboot using Google Gemini with privacy-centric design. Plus: what Nvidia earnings could signal for the AI infrastructure boom, why memory and storage are emerging bottlenecks, how agentic AI is compressing high-skill finance work, where measurable ROI is showing up in hiring, why sovereign AI is reshaping enterprise architecture, and how synthetic media is becoming a mainstream brand-safety issue. In AI in Flow, the message is clear: capability matters, but control, trust, and infrastructure matter more.

Claire and Peter break down the day’s biggest shifts shaping AI beyond the hype. A Gallup survey finds 71% of Americans oppose data centres near their homes, turning “the cloud” into a local politics and infrastructure battle. They explore warnings that the high-bandwidth memory boom could face a familiar boom-bust cycle, and why demand could cool if models get more memory-efficient. Plus: the surge in forward deployed engineers as companies scramble for people who can ship AI into real operations; Anthropic’s Claude Cowork and the governance stakes of desktop agents that can act across apps and files; China’s move toward a comprehensive AI law; AI’s growing presence in film and the need for transparent disclosure; Merriam-Webster naming “AI slop” Word of the Year; capital rotating toward AI-heavy chip markets; how AI may reshape work more than remove it; and the often-overlooked materials supply chain—copper and specialty metals—powering the entire buildout. All on AI in Flow.