🧠 Your next coworker won’t be human, and it might already be preparing your next meeting. Or your demise. French startup Dust believes the future of work isn’t about AI assistants but about teams of humans and AI agents working side by side. We spoke to co-founder Stanislas Polu about what’s coming next.

Chris O'Brien + Helen O'Reilly-Durand

Headlines

🗞️ Sifted's Daphné Leprince-Ringuet has the scoop on a new stealth AI startup with deep roots at OpenAI and Meta that is aiming to make a very big entrance. According to Sifted, Aire (registered as FLOPS & FRIENDS) is seeking $400 million in Seed funding at a reported valuation of around $2 billion to build an "AI scientist" capable of automating R&D for enterprise customers in fields ranging from biology and physics to finance. The company was co-founded by former OpenAI researcher Irwan Bello alongside ex-Meta FAIR researchers Gabriel Synnaeve, who is based in Paris, and Yossi Adi, with operations spanning the US, France and Israel. If the raise succeeds, Aire would join a growing wave of heavily funded AI labs betting that the next frontier for AI is accelerating scientific discovery rather than simply boosting office productivity. | Sifted

🗞️ French ride-sharing giant BlaBlaCar is hitting the accelerator, using AI to power an ambitious expansion into 20 new markets. By leveraging AI for operational scaling, translation, and localized advertising, the Paris-based company has achieved in mere months what previously took years with a traditional rollout. It’s a classic case of using smart tech to streamline growth, helping the platform expand its footprint across regions such as Southeast Asia, Latin America, and North Africa. | LinkedIn

🗞️ The deadly earthquakes in Venezuela have triggered a wave of AI-generated and recycled videos spreading across social media, complicating rescue efforts and public understanding of the disaster. Fact-checkers have identified three dominant forms of visual misinformation: footage recycled from disasters in other countries, old videos falsely presented as current, and increasingly convincing AI-generated clips depicting building collapses that never happened. The episode underscores how generative AI is amplifying crisis misinformation and making real-time verification more critical than ever. | France 24

🗞️ Reminder: The countdown is on for businesses ahead of the EU AI Act’s next major milestone on August 2. While some high-risk AI rules have been postponed, transparency obligations will still come into force for many common AI applications, including chatbots, AI-generated content and synthetic media. Experts say companies should now move beyond simply tracking the regulation and start auditing where AI is already being used across their organizations—including “shadow AI” embedded in everyday software. | Actuia

🗞️ France has passed a scaled-back anti-fast-fashion law targeting ultra-fast retailers, including Shein, Temu, and AliExpress. The legislation introduces environmental fees on qualifying garments and bans advertising by ultra-fast fashion brands, but stops short of applying the rules to European retailers such as Zara and H&M, drawing criticism from campaigners. While the law focuses on environmental impact rather than AI, it chiefly targets companies like Shein, whose AI-powered algorithms drive ultra-fast product design, demand forecasting and personalized shopping at massive scale. | Le Monde, France24

Check out the France x Shein back story here

🗞️ French healthtech leaders, including Alan, Doctolib, Lifen, Implicity and Resilience Care, have urged the government to give the sector a stronger voice in shaping France’s sovereign cloud strategy. French healthtech leaders, including Alan, Doctolib, Lifen, Implicity and Resilience Care, have urged the government to give the sector a stronger voice in shaping France’s sovereign cloud strategy. | Les Échos

🗞️ France’s sovereign AI assistant is reducing civil servants’ reliance on ChatGPT, but not replacing it. A six-month pilot involving 10,000 public sector employees found that 75% considered the government’s chatbot useful, and more than half said they now use tools like ChatGPT less often at work. Yet 57% still believe leading U.S. AI assistants deliver better responses, highlighting the challenge of balancing digital sovereignty with user expectations... | L’Usine Digitale

🗞️ French e-signature platform Yousign (soon to become Youtrust) now lets customers manage documents directly from AI assistants including ChatGPT, Claude and Mistral’s Vibe. Users can send signature requests, track document status, and check whether contracts have been signed without opening the app. The company says AI assistants are rapidly becoming the new interface for business software, with integrations for Gemini and Harvey already on the roadmap. | L’Usine Digitale

🗞️ French cybersecurity firm YesWeHack has launched an AI-powered “agentic pentest” platform that cuts penetration testing from five days to just one. Rather than relying on a single large language model, the system orchestrates multiple AI agents to mimic the reasoning of a human ethical hacker, identifying and validating exploitable vulnerabilities before producing remediation recommendations. The company says the technology is designed to accelerate—not replace—its community of 150,000 ethical hackers. | L’Usine Digitale

🗞️ Digital twins are evolving from virtual replicas into AI-powered operating systems for critical infrastructure. As AI agents, sensors, drones, and edge computing become tightly integrated, digital twins are increasingly used not just to visualize assets but also to simulate failures, orchestrate maintenance, and support real-time operational decisions. From Orange’s network demonstrations at VivaTech to Europe’s Destination Earth initiative, the technology is moving beyond dashboards toward managing the physical world. | Actuia

🗞️ AI may be fueling the next wave of IT spending, but it’s also exposing how much infrastructure goes to waste. French startup Sopht says nearly a third of cloud spending is wasted and is helping enterprises optimize the cost and environmental impact of their IT as AI workloads continue to grow. The company has just raised €7.5 million to expand across Europe. | Maddyness

🗞️ Paris’ Raise Summit is fast becoming one of Europe’s flagship AI events. Ahead of its third edition next week, organizers say attendance has grown from around 2,000 in 2024 to an expected 9,000 this year, with speakers from OpenAI, Anthropic, Google, Mistral AI and Cognition. The conference has expanded with the launch of Machina, a new event dedicated to AI-powered robotics, reflecting growing interest in “physical AI.” | Maddyness

🗣️Responsible AI🗣️

A look at the big stories this past week from Responsible & Frugal AI expert James Martin, of BetterTech.

Is AI a bubble? A surprising source has replied: ‘possibly’. The Bank for International Settlements - which advises global central banks, no less - is concerned about an “investment bust” caused by “AI exuberance,” says the FT.

Market uncertainty could be China’s opportunity, especially with Anthropic and now OpenAI’s latest models restricted by the Trump administration. Zhipu revealed GLM 5.2, a model that rivals Opus 4.8, according to CNBC, but at a fifth of the cost… and 2-3 times smaller, meaning you don’t need cutting-edge hardware to run it. Dspark, new tech from DeepSeek that improves response speed by 85% and reduces the need for the latest GPUs, according to SCMP. And of course, Chinese models are open source (enough) to host wherever you like.

Just don’t ask them about Taiwan or Tiananmen Square. France’s finance ministry just removed a Qwen model from an experiment aimed at helping (human) agents in the field, as it demonstrated ingrained pro-China bias. Mistral AI swooped in to replace it, reports Le Monde. Cocorico!

Meanwhile, those hoping AI will transform the workplace for the better may be disappointed : 

• CEOs are increasingly mistaking ChatGPT or Claude for their “vizirs” (think Jafar in Aladdin), reports Futurism, taking their often-sycophantic advice on decisions as crucial as whom to fire, bonuses, or new company strategy ‘ideas’
• Workers fired ‘because of AI’ are getting called back to work after said systems’ limitations become evident: Ford just rehired 350 veteran engineers for that reason, reports The Independent
• Meta’s cunning plan to replace its workers with agents by tracking their every mouse and on-screen movement has already backfired… but not because it’s illegal (in Europe); rather, because Meta let data harvested so far become available to all of its staff, reports Business Insider. Oopsie!

At least, at last, the UN is taking on AI’s environmental impacts. Secretary General Antonio Guterres reminded us that by 2030, AI infrastructure could consume enough water to cover the needs of all 1.3 billion residents of Sub-Saharan Africa for an entire year. Whence his call for “every major AI company to measure and publicly disclose the full environmental impact of its systems.” 

Might this ever happen? Maybe with enough pressure from both consumers and clients. As a former AWS sustainability official told AP, AI providers will “listen if everybody suddenly starts caring about not having a footprint.” 

That is, of course, if peaceful data center protestors aren’t arrested for no reason, or have their anti-AI lawsuits threatened by the US Department of Justice…

🧠 Your Next Coworker Won’t Be Human. Dust Thinks That’s Just the Beginning

On Monday morning, an AI agent logs into a company’s systems. It pulls sales figures from HubSpot, product updates from GitHub, and financial metrics from Snowflake. It prepares the weekly management presentation and then sends messages to eight team leaders, asking them to review or update their respective sections.

By Tuesday lunchtime, the presentation is complete and waiting for the manager responsible for the meeting to sign off on it. No one spent hours chasing colleagues for slides. The AI agent did the coordination.

Welcome to the Brave New World of Dust.

☀️ Sponsor Events☀️

📆 MACHINA | July 7 | Paris | The international summit dedicated to humanoids, robotics, industrial autonomy, and next-generation human-machine interfaces. | Buy Tickets

📆 RAISE Summit 2026 | July 8–9, 2026 | Carrousel du Louvre, Paris | Join one of Europe’s most influential AI gatherings bringing together thousands of builders, investors, executives, and innovators to shape the future of artificial intelligence. Learn from global leaders through keynotes, panels, workshops, a hackathon, startup competitions, and high-impact networking — all curated to accelerate AI strategy, investment, and real-world deployment across industries. | Tickets

📆 Events 📆

📆 Economic Data Science Society (EDSS) Conference 2026 | July 9–10 | Banque de France, Paris | The inaugural conference at the intersection of AI, data science, and economics, bringing together researchers, policymakers, and industry leaders from institutions like the IMF, ECB, and leading universities. Explore topics such as machine learning for economic forecasting, LLMs in economic analysis, alternative data (satellite, web scraping), and AI governance — alongside keynote talks, academic sessions, and hands-on training workshops. | Apply

📆 The Agentic GTM Hackathon by Anthropic, FullEnrich & Sillage | July 9 | Station F, Paris | Anthropic, FullEnrich, and Sillage invite the sharpest builders, sellers, and GTM engineers for a full-day hackathon to answer one question: what does AI-native revenue look like when you actually build it? | Request to Join

👋🏻 If you’re enjoying The French Tech Journal, support the project by forwarding it to friends and sharing it on your social networks. You can also comment on this post. And if you have ideas for stories, tips, or just want to harass us, shoot us an email: chris@frenchtechjournal.com / helen@frenchtechjournal.com 👋🏻‌