Quantum IPO Records, Advanced AI Safeguards, and Public Sector Transformations

QUANTUM COMPUTING : IPOS AND FUNDING INITIATIVES FLOURISH

Quantinuum’s IPO on the week of June 1st raised $1.68B in the largest quantum computing IPO in history, the latest milestone in a dramatically accelerating market. Quantinuun is the result of the merge between the UK Cambridge Quantum and the US Honeywell Quantum Solutions.

At the same time, the Finnish IQM finalizes its IPO, for an initial valuation of 1,8 B€. A

nd the French QUOBLY secures a new funding of 128 M€.

AI 

ANTHROPIC ISSUES THE MOST ADVANCED MODELS AND LIMITS THEIR USE

Claude Fable 5 is directly derived from the Mythos family. The model steps up a notch in terms of benchmark performance and becomes the smartest model released to date by Anthropic. The model performs exceptionally well in software engineering tasks, knowledge work (advanced reasoning ability), and vision (analysis of visuals and images). Anthropic describes it as superior to its competitors’ frontier models on complex tasks that require high autonomy. But such powerful model also means higher than normal risk. To prevent their model from being misused by malicious actors, Anthropic has developed new guardrails specifically tailored for the intelligence of Fable 5.

A direct consequence of this fear, the safeguards of Fable 5 prevent the model from responding to all topics related to biology, chemistry, model distillation, and cybersecurity. And Sam Altman (OpenAI), Dario Amodei (Anthropic), Demis Hassabis (Google DeepMind), Mustafa Suleyman (Microsoft AI), and Alexandr Wang (Meta) have co-signed a public letter calling for mandatory regulation of the sale of synthetic DNA and RNA in order to prevent AI models from being used to create biological weapons.

FRANCE TRAVAIL REPORTS BETTER SERVICE TO UNEMPLOYED PEOPLE THANKS TO AI

AI at FRANCE TRAVAIL allows both to relieve advisors of administrative tasks and to better support job seekers in their search. ChatFT is a conversational AI that helps advisors write job postings and emails or summarize interviews with job seekers. With ChatFT Listen, a version of the chatbot, the advisor no longer has to worry about filling in the fields of their information sheet during exchanges with a job seeker. The AI transcribes their conversation into useful data to feed the information system. This new tool is being tested in six regions, including Centre Val de Loire, Aquitaine, and Hauts de France, before a wider rollout in 2027.

As its name suggests, MatchFT is a “matching” tool aimed at reconciling supply and demand. When presented with a recruiter’s need, France Travail doesn’t always have all the useful and up-to-date information to know if a candidate can meet it. Is the candidate available from 7 a.m.? Do they have a car?

The AI sends a series of texts to check the job seeker’s interest, availability, or mobility. On average, the job seeker responds in less than an hour and a half. The average time for France Travail to fill a position drops by 3 days.

Share:

More Posts