Tech Brief - EU Commission briefing book leaked, CSAM COREPER vote cancelled

“In line with the objective of simplification and single market focused as included in the High-Level Report on the future of the Single Market by Enrico Letta, regulatory uncertainty and regulatory burden should be reduced.”

– The EU Commission in its briefing book on digital.

Story of the week: The next European Commission should have a mid-term review of its digital rulebook and revise the Audiovisual Media Services Directive (AVMSD), says an internal briefing document seen by Euractiv. The document, likely dating back to January, largely focuses on implementing the legislation passed in the previous mandate. The next Commission commits itself to a mid-term review of the digital rulebook, such as regulation on competition, the Digital Markets Act, content moderation, the Digital Services Act, and artificial intelligence. Read more.

Don’t miss: The draft law to detect and remove online child sexual abuse material (CSAM) was removed from the agenda of Thursday’s meeting of the Committee of Permanent Representatives (COREPER), who were supposed to vote on it. The vote was likely removed due to disagreements about the regulation, which probably would not have secured a sufficient majority for it to pass. While there are still some meetings scheduled until next month, the regulation is unlikely to reach COREPER again until Hungary takes over the EU Council presidency on 1 July. Read more.

Also this week:

Before we start: If you just can’t get enough tech analysis, tune in on our weekly podcast.

Final step for AI Factories. The Council of the EU adopted an amendment to the European High-Performance Computing (EuroHPC) undertaking on Wednesday. This clears the way for creating AI factories, supercomputers that will be made available to “innovative startups and SMEs.”

AI Board. On Wednesday, the Commission hosted the first high-level meeting for the upcoming EU AI Board to prepare for the AI Act’s implementation, expected in early August. Key attendees included high-level delegates from EU Member States, the European Data Protection Supervisor, and EEA/EFTA representatives. The meeting aimed to ensure a robust and timely AI governance framework. The next meeting is scheduled for early autumn.

Masayoshi is back. SoftBank’s founder Masayoshi Son told shareholders he is ready to make a big move, hinting at a big AI investment, on Friday, The Japan Times reported on Thursday. Son has shied away from the spotlight since his company lost billions in the WeWork saga, followed by a string of bad investments. He has instead focused on SoftBank’s chipmaking unit, Arm Holdings. SoftBank announced a partnership with a generative AI startup, Perplexity AI, earlier this week. The startup has been accused of ripping off Forbes’s work.

OpenAI defector launches AI startup. Co-founder of OpenAI Ilya Sutskever, who left the company a month ago, is launching his own startup focused on building “safe superintelligence,” he said on X on Wednesday. The co-founders are another OpenAI alumni, Daniel Levy and Daniel Gross, who worked at startup accelerator Y Combinator, which OpenAI CEO Sam Altman used to run.

New AI model on the market. Anthropic, supported by Google and Amazon, unveiled a new artificial intelligence (AI) model and a new interface on Thursday to boost user productivity. Claude 3.5 comes three months after its Claude 3 series of AI models.

Hungarian Presidency on cybersecurity. The Hungarian Presidency’s six-month agenda was published on Tuesday. It included a possible review of the Cybersecurity Act as well as a revision of the 2017 Commission recommendation on coordinated response to large-scale cybersecurity incidents.

ENISA tests energy sector. ENISA challenged the cyber-resilience of the EU energy sector in the seventh edition of its annual test, called “Cyber Europe.” Representatives from Ukraine and the US were at ENISA headquarters in Athens, where the exercise was managed from. The energy sector ranks among the lowest among critical sectors in cyber resilience, the agency said.

French cybersecurity agency warns against Russian cyberattacks on diplomatic entities. The ANSSI published a report on Wednesday explaining that “several cyberattacks against French diplomatic entities can be linked” to Russian hacker group Nobelium. The report further notes that “several IT companies” also reported incidents linked to Nobelium in 2023 and 2024.

No AI for Meta users in Europe. Facebook and Instagram’s parent company Meta is pausing plans to roll out AI tools in Europe, following a “request” from Ireland’s Data Protection Commission (DPC), the firm said last Friday. Meta’s “decision followed intensive engagement between the DPC and Meta,” said the Irish authority. Digital rights NGO filed complaints. The company said it is “disappointed” by the Irish DPC “request”, which came “on behalf of” European DPAs “to delay training our large language models (LLMs) using public content shared by adults on Facebook and Instagram.” Read more.

Apple and OpenAI’s saga could raise eyebrows. The new Apple and OpenAI partnership, announced last week, will integrate ChatGPT into Apple’s operating systems, virtual assistant, and writing tools. It poses important questions about competition and privacy, experts told Euractiv. Apple’s integration of ChatGPT could add to regulatory scrutiny about competition, including the risk of collusion among competitors sharing partnerships with the same AI firm. There are also worries about the impact on third-party app access and user privacy and processing shifts from on-device to cloud-based models, prompting scrutiny under the EU’s Digital Markets Act and data protection laws. Read more.

Amazon to invest €10 billion in Germany. On Wednesday, Amazon announced a plan to invest €10 billion across “100 German cities” in cloud and logistics. In May, Amazon announced a €7.8 billion investment in what it calls “sovereign cloud” in Germany.

Meta and AI audio. Meta’s AudioSeal embeds hidden watermarks in AI-generated audio to combat scams and misinformation, achieving high detection accuracy, MIT Technology Review reported on Tuesday. It can identify AI-generated segments in long clips and remains detectable even after editing. However, challenges include the lack of industry standards, ease of tampering, and the need for voluntary adoption. Despite its potential, scepticism remains about its robustness against attacks and its impact on public trust.

TikTok faces another US complaint. The US Federal Trade Commission referred a complaint against TikTok and its parent company ByteDance to the Department of Justice on Tuesday. The complaint is about potential violations of the US Children’s Online Privacy Protection Act.

Hungarian telecom agenda. During its presidency, Hungary will look to adopt Council conclusions on the Commission’s white paper on the future of telecommunications while reviewing existing legislation, including the EU Electronics Communications Code.

Starlink in France. The French regulatory authority Arcep announced on Wednesday it has opened a public consultation until 19 July on Elon Musk’s Starlink request to modify its spectrum bandwidth usage.

Indonesian telecoms vs. Musk. Just weeks after launching Starlink in Indonesia, local internet services providers are calling on the government to halt the sale of its services, Nikkei Asia reported on Wednesday. The providers argue that Starlink could disrupt the local industry and that regulation should be passed. Meanwhile, the government is courting Musk for other investments, such as a SpaceX rocket launch site and a Tesla electric car factory.

[Edited by Alice Taylor]

Thanks for reading. Be sure to spread the word and come and say hello on X.