On September 3, 2026, Anne Le Hénanff, minister delegate in charge of Artificial Intelligence and Digital Affairs, brought together at Bercy the stakeholders of the “Osez l’IA” plan to mark its first anniversary. The press release and the article published on economie.gouv.fr the following day restate almost word for word the statement under review: launched on July 1, 2025 as part of the national AI strategy, the plan aims to accelerate the spread of artificial intelligence in companies to strengthen competitiveness and stimulate innovation. It presents a “first positive assessment” with “first concrete results”. Established facts and origin of the buzz. The July 1, 2025 launch is confirmed by multiple official and contemporary press sources (Bpifrance, Clubic, Silicon, ActuIA). The plan is structured around three pillars: raise awareness (ambassador network), train (AI Academy) and support (diagnostics, catalogue, accelerators, funding via France 2030). The 2030 objectives remain unchanged: 100% of large companies, 80% of SMEs and mid-sized firms, 50% of very small businesses using AI. Originally, the network targeted 300 ambassadors and nearly 5,000 co-funded diagnostics; the overall envelope was around 200 million euros, part of which was already earmarked in France 2030. The one-year assessment figures, taken up in press release No. 990 of September 3, 2026 and on the ministry’s website, are as follows: more than 35,000 companies made aware thanks to 615 AI Ambassadors spread across 20 regions and local authorities (including overseas territories) and 13 sectors; 88 AI solution providers suited to SMEs/mid-sized firms listed in a DGE/Hub France IA catalogue; 70 Data AI Diagnostics initiated; “Accelerate with AI” platform (launched at Vivatech in June 2026) bringing together 54 use cases. A new stage is announced to move from awareness-raising to action: stepping up field work, enriching the platform and catalogue, rolling out 1,000 Data AI Diagnostics (an 8 person-day service at around €10,000 excluding tax, 40% covered by France 2030), and new cohorts of Bpifrance AI Accelerators (18 months, 46% covered). The government also states that AI adoption by companies has “tripled in two years”. These plan activity elements are consistent across the economie.gouv.fr website, the press release and the pages of the Directorate General for Enterprises. They were announced progressively, notably a reinforcement in June 2026 at Vivatech (rise to 615 ambassadors). What is accurate, what is overstated or still uncertain. The launch date, the objectives, the plan’s architecture and the activity figures (awareness-raising, ambassadors, diagnostics initiated, catalogue, use cases) match official communications. There is no major internal contradiction. By contrast, describing the assessment as “positive” and the results as “concrete in companies” is a matter of government appraisal. The specialized press offers a more measured view. LeMagIT headlines “a modest start” on September 4, 2026: 70 diagnostics against a stated target of 1,000, 54 use cases to cover the entire economic fabric, and the figure of 35,000 companies made aware already announced in June. L’Usine Digitale, the same day, stresses that the State has “mainly raised companies’ awareness, without yet demonstrating their transformation” and that the road remains long toward the 2030 targets. The claim of a tripling of adoption in two years draws on general trends, not solely on the plan. Insee (ICT companies survey, Insee Première No. 2120, July 2026) indicates that in 2025, 18% of companies with 10 or more employees used at least one AI technology, versus 6% in 2023 (a threefold increase) and 10% in 2024. The rate reaches 58% for firms with 250 or more employees, but only 15% for those with 10-49 employees. Sectoral and size gaps remain marked. A Banque de France survey (early 2026, about 7,000 companies with 20 or more employees, published around September 3, 2026) shows rapid diffusion of generative AI: 67% use it (up to 83% for those with more than 1,000 employees), often via consumer tools. However, only a minority already report productivity gains (around 31% for generative AI), even though optimism for the next three years remains high. Les Echos and BFM relay these findings: growing adoption, but still limited productive impact and uses that are often experimental or concentrated on support functions. It is difficult to isolate the causal effect of the “Osez l’IA” plan from the broader dynamic (ChatGPT and similar tools, competitive pressure, other France 2030 schemes). The 35,000 awareness actions represent real field activity via the ambassadors, but do not measure the rate of transition to lasting implementation. The 70 diagnostics initiated remain a low volume relative to the initial ambitions and to the number of French SMEs/mid-sized firms. The catalogue of 88 providers and the use-case platform are useful tools, but their actual use and their impact on competitiveness have not yet been documented independently at scale. Contradictions, blind spots and context. The plan has evolved: the ambassador network has more than doubled the initial target of 300. Diagnostics have moved from ambitions of several thousand to a first milestone of 1,000. Training figures (target of 15 million professionals trained by 2030 via the AI Academy) are not detailed in the one-year assessment. No independent public data precisely quantifies, as of September 2026, the plan’s net contribution to national adoption rates. Structural barriers (lack of expertise, absence of identified use cases, costs, data maturity) remain cited in the Insee and Banque de France surveys. The presumption of innocence and privacy are not at issue here; this is a matter of public policy. On X (formerly Twitter), posts around September 3-5, 2026 mainly relay the ministry’s official press release and share the critical articles from L’Usine Digitale and LeMagIT. Few heated debates; a few mentions of ambassadors present at Bercy. Provisional conclusion. The official statement is factually accurate on the launch, the objectives and the plan’s activity indicators after one year. The “positive assessment” and the “concrete results” correspond to government communication and to a ramping-up of awareness-raising and tooling schemes. However, in light of the still low volumes of in-depth support (70 diagnostics) and independent press analyses, the effective shift to massive and transformative adoption remains to be confirmed. AI adoption is progressing markedly in France for broader reasons, but productivity gains are still awaited and the 2030 objectives remain ambitious. A finer independent evaluation of the plan’s causal impact, beyond activity figures, would be needed to fully judge its effectiveness. The verdict on the “positive” nature of the first assessment therefore remains open and proportionate to the available evidence: solid operational starts, still limited transformation results.