France inaugurates INESIA, its IA / DOGE security center, the silent intrusion of Musk / Openai engineers launches Deep Research

With our partner Salesforce, unify sales, marketing and customer service. Accele your growth!

France inaugurates Inesia, its IA security center

The French government officially launched the National Institute for the Evaluation and Security of AI (Inesia) on January 31. This new institute aims to produce tools to assess and improve the performance of AI models, as well as to detect the systemic risks posed by advanced technologies.

Budgetary rigor obliges, Inesia will operate without budget or dedicated human resources, and will have to count on the collaboration of several administrations and agencies of the State, such as Inria, ANSSI, and LNE. Inesia financing will come from existing programs, including an investment of 20 million euros awarded by the France 2030 program to INRIA.



We could not confirm your registration.


Your registration is confirmed.


A state under influence: the silent intrusion of Musk engineers

The arrival of a group of engineers, aged 19 to 24 and mainly from Elon Musk companies, in key roles in the federal administration raises questions about transparency and security of government infrastructure. Placed within the Department of Government Efficiency (DOGE), these engineers have obtained privileged access to strategic agencies such as the Office of Personal Management (OPM) and the General Services Administration (GSA), without going through the classic processes of appointment and security verification.

Their presence is part of an initiative that officially aimed at modernizing the government’s digital systems, but several incidents have strengthened concerns. Attempts at unauthorized access to classified data have been reported, in particular to the United States agency for international development (USAID), where security officials who blocked the intrusion were then put on forced leave .

The fears do not relate only to the inexperience of engineers, but to the lack of institutional control around their mission. Unlike conventional officials, these actors have not been appointed via standard procedures, which complicates any parliamentary supervision. The question then arises: are we witnessing an accelerated modernization of the administration, or a discreet takeover of critical infrastructure by a private player?

The lack of official responses from the agencies concerned fueling doubts about the real nature of their role and the risks of capturing government data. The stake exceeds the simple presence of these engineers: it affects the capacity of the State to guarantee its independence in the face of private interests.

The United States excludes works generated by copyright

THE US Copyright Office confirmed that the works produced only by artificial intelligence cannot be protected by copyright. Only one perceptible human contribution makes it possible to establish legal paternity.

The Office cites the example of ROSE ENIGMAan illustration where the artist had drawn a sketch before using an AI to enrich the details. The Copyright Office granted protection uniquely On the elements of the original drawing visible in the final image. Conversely, Suryasta photo modified by the AI Raghav To imitate the style of Starry Night of Van Gogh, was rejected, the transformation being judged fully automated.

On promptthe office tested a request requiring a “cat with glasses reading a newspaper”. The results varied: some displayed a cat with realistic features, others a feline holding the newspaper with a human hand. These differences demonstrate that the user does not control directly The final expression, which excludes any claim from author.

The report concludes that the current law is sufficient and Reforms any reform project. While China and Japan are exploring partial recognition of AI assisted works, the United States maintains a strict position: Only a human can create and claim their copyright

Softbank won 3 billion per year for its own use of OpenAi technologies

As part of this collaboration, the two entities also revealed the creation of “SB Openai Japan”, a joint venture that will exclusively market Openai business technology with large Japanese companies.

This initiative will offer SoftBank and its subsidiaries access to products such as Chatgpt Enterprise, Openai API, personalized models and agent products like Operator. Arm, the British flea designer acquired by SoftBank in 2016, will also use OpenAi tools to “improve productivity within the company”.

During an event broadcast live on Monday morning, organized jointly by Softbank, SoftBank Group, Openai and Arm, the leaders present represented companies whose cumulative market capitalization constituted more than half of that of the Japanese market. Masayoshi Son, speaking through a translator, shared his conviction that general artificial intelligence (AG) will become a reality in less than ten years, probably starting with large companies thanks to the substantial investments available only in this type of structures.

Geospy: AI capable of locating you from an image detail

Geospy, developed by Graylark Technologies, uses AI to geolocate a photo in a few seconds, without metadata. Its algorithm analyzes architecture, vegetation and decoration elements to identify a place with impressive precision. First used by the police, the tool was diverted for surveillance and harassment, pushing its publisher to restrict access.

OPENAI launches Deep Research, a new tool for internet research

OPENAI today announces the launch of Deep Research in Chatgpt, a new solution intended to carry out internet research for complex tasks. Deep Research is able to carry out research in a few minutes that would take hours to a human. This tool represents a major advance towards the wider objective of AGS development (artificial general intelligence), a form of AI capable of producing sharp scientific research.

The United Kingdom hardens legislation against images of sexual violence on children generated by IA

The British government announces new laws to combat images of sexual violence on children produced by artificial intelligence. It will now be illegal to own, create or distribute IA tools to generate such images, with sentences of up to five years in prison. The possession of IA pedophile manuals will also be prohibited, with up to three years in prison for offenders. Yvette Cooper, Minister of the Interior, underlines the urgency of these measures in the face of the increase in online abuse amplified by the AI.

Intel obtains a reimbursement of 1.06 billion euros and the late interest on the part of the EU

In October 2024, the Court of Justice of the European Union ruled in favor of Intel, resulting in the reimbursement of a fine of 1.06 billion euros, initially inflicted in 2009. Following this decision, Intel also received delay interests amounting to 515.55 million euros.

+ Discover

What is a CEP (Customer Commitment Platform)?

+ Experiences

Spin Selling: a methodology that transforms the art of asking the right questions

+ Fundraising

Riot lifts $ 30 million to counter cyber attacks doped at AI

  • ThePackergers raises 14 million euros to industrialize the logistics of e-commerce from art and vintage
  • Zaion raises 11 million euros to transform banking customer relations with AI
  • Embedded Finance is essential: Swan raises 42 million euros to extend its model in Europe
  • Trade raises 21 million euros to build the new generation open-source financial infrastructure
  • Cryptio raises an additional € 15 million for its financial solution in the crypto

Trends:

Geospy: AI capable of locating you from a simple detail on an image

To register in your agenda:

Tech & Fest, the new flagship festival of innovation not to be missed and it’s February 5 and 6, 2025

The job offer with Altaide Recruitment Cabinet

Alan recruits his CEO Founder Associate for a strategic internship in 2025

Alan opens a unique opportunity for a trainee at the end of studies or in the year of caesura : the post of CEO Founder Associateto be filled between June and December 2025 for a period of six months. This internship offers a strategic immersion at the heart of the company’s priorities, under the direct supervision of the CEO and the Alan managers.

The mission includes: strategic analyzes, coordination of transversal projects, management of new products, or support for tenders and fundraising.

🥐 previous edition of the Good Morning Frenchweb:

    • But where did the report of the AI ​​commission go? / AI2 defies Deepseek and Openai with tulu 3 405b / Mistral AI accelerates the cadence
    • “Competitiveness COMPASS”: the EU in the face of its contradictions / Bending Spoons: the Italian startup which was worth 3 billion euros / Swan lifts 42 million euros …
    • Who is Liang Wenfeng, the man behind Deepseek? / Google Maps reclassifies the US as “sensitive country”/ 4 fundraisers share 130 million euros.

🚀 To read on Frenchweb.fr:

    • Bending Spoons: the Italian startup that was worth 3 billion euros
    • Associates Pact: What clauses are the most problematic when selling a startup?
    • Summit for Action on AI: Europe suffocated by its own rules?