☕️ Nobel laureate leaves DeepMind for Anthropic

Hi there, this is your daily ☕️ Techpresso.

In today's Techpresso:

🏆 Nobel laureate leaves DeepMind for Anthropic

🤝 Trump says Anthropic is no longer a national security threat

🇳🇴 Norway bans AI in elementary schools

🛡️ Meta wants child safety bill rewritten to dodge lawsuits

🤖 VLC creator now wants to control robots

Plus: 🎁 5 other news you might like, 🧰 6 tools, and 📚 3 papers.

🏆 Nobel laureate leaves DeepMind for Anthropic LINK

John Jumper, who won the 2024 Nobel Prize in chemistry for the AlphaFold protein-prediction model, is leaving Google DeepMind after nearly nine years to join Anthropic, the maker of Claude, both companies confirmed Friday.

Jumper announced the move on X, crediting DeepMind chief Demis Hassabis for letting him lead the AlphaFold team six months after his PhD, and said he would take time off before starting his new role.

Neither Jumper nor Anthropic disclosed his role, though the hire fits the company's push into life sciences, following its $400 million stock purchase of biotech startup Coefficient Bio in April.

🤝 Trump says Anthropic is no longer a national security threat LINK

President Trump said in an Axios interview published June 19 that he no longer sees Anthropic as a national security threat, two days after meeting CEO Dario Amodei at the G7 summit in France.

The shift followed a June 17 G7 working lunch where Amodei and Demis Hassabis pitched a US-led AI coalition, though the Commerce Department's June 12 order restricting foreign access to Mythos 5 and Fable 5 stays in force.

Trump said the original alarm came from a competitor that also owns Anthropic, matching Amazon, which detailed a vulnerability in Mythos and brought it to the administration before Commerce Secretary Lutnick sent the export-control letter.

🇳🇴 Norway bans AI in elementary schools LINK

Norway is set to bar children between the ages of 6 and 13 from using AI in elementary schools, with the near-ban taking effect in late August across the country.

Teenagers between 14 and 16 will only be allowed to use AI under direct teacher supervision, while students aged 17 to 19 will learn to "use AI appropriately" for further education and work.

Prime Minister Jonas Gahr Støre said using AI raises the risk that young children miss important steps in learning to read, write, and do mathematics, and the government will also propose legislation promoting books in classrooms.

🛡️ Meta wants child safety bill rewritten to dodge lawsuits LINK

Meta has lobbied Congress to add a provision to the Kids Online Safety Act that would grant platforms immunity from state-level child-harm lawsuits involving users under 18, possibly undercutting thousands of cases already filed.

The proposed language would make companies immune from state-law claims tied to the safety or privacy of minors online, and it would also override certain state laws governing children's online protections.

According to a source, Meta offered to drop its opposition to KOSA if the provision is included, while Senator Marsha Blackburn's office said it had not seen the language and would not support it.

🤖 VLC creator now wants to control robots LINK

Jean-Baptiste Kempf, the lead developer of VLC Media Player, has raised $5 million for his startup Kyber, an infrastructure layer built to control remote devices like robots and drones in real time.

Kyber's core product is an SDK that synchronises video, audio, sensor data, and control inputs, and Kempf showed it reaching 8 milliseconds of glass-to-glass latency, building the platform on top of FFmpeg and VLC.

The Paris-based company, already deployed with customers in defence, telco, robotics, and AI, is prioritising robotics, drones, and remote IT access, where Kempf positions it as a challenger to Citrix.

Other news you might like

• Encryption, spyware, and now Mythos: History shows why cyber export control doesn’t workLINK

• This month's Windows 11 update broke the Recycle Bin, OneDrive, and possibly your PC's stabilityLINK

• Ubisoft Co-Founder Claude Guillemot Dead After Plane Crash In FranceLINK

• ChatGPT keeps creeping toward becoming your AI personal assistant with new scheduled task controlsLINK

🧰 Trending tools

WorkClaw: collaborative AI coworkers that integrate into Slack and Teams, manage tasks across 3,000+ apps, and work autonomously with your entire team 24/7.LINK

Slackbot's MCP Client: connects Slack's messaging and collaboration features to MCP-compatible tools, letting teams trigger workflows and share data without leaving their workspace.LINK

Foyer: turns any Notion page into a waitlist with referral rewards, generating a shareable signup URL in seconds without extra tools.LINK

Mellum by JetBrains: a code-focused LLM built for IDE integration, helping developers autocomplete code and generate context-aware suggestions faster.LINK

pumaDB: a lightweight shared memory layer for AI agents, letting them persist and reuse context across sessions without database setup or RAG infrastructure.LINK

ReleaseDock: combines support inbox, AI agent, help center, changelog, and embeddable widget so teams spend less time on support infrastructure.LINK

📚 Trending papers & reports

Loopholes in social rules can be automatically discovered by reward-trained chatbots, which learn to follow laws technically while breaking their intent, and current safety guardrails offer only limited protection.LINK

GitHub Copilot makes engineers complete 40.5% more pull requests, or finished chunks of code, in their heaviest usage weeks compared to weeks they don't use it at all, holding total effort constant.LINK

Measuring intelligence now has a rigorous universal formula, showing that any truly intelligent system must simulate itself and its own future actions to meaningfully change what happens next.LINK

Want to get the latest news differently? Find us on:

See you tomorrow for a new dose of ☕️ Techpresso!