Close Menu
TechTost
  • AI
  • Apps
  • Crypto
  • Fintech
  • Hardware
  • Media & Entertainment
  • Security
  • Startups
  • Transportation
  • Venture
  • Recommended Essentials
What's Hot

Gradient’s heat pumps get new smarts to enable retrofitting of old buildings

Peak XV Says Internal Disagreement Has Led to Partner Exits as AI Doubles

New York lawmakers are proposing a three-year freeze on new data centers

Facebook X (Twitter) Instagram
  • About Us
  • Contact Us
  • Privacy Policy
  • Terms and Conditions
  • Disclaimer
Facebook X (Twitter) Instagram
TechTost
Subscribe Now
  • AI

    New York lawmakers are proposing a three-year freeze on new data centers

    7 February 2026

    Benchmark raises $225 million in dedicated funds to double Cerebras

    7 February 2026

    How artificial intelligence is helping to solve the labor issue in treating rare diseases

    6 February 2026

    Amazon and Google are winning the AI ​​capital race — but what’s the prize?

    6 February 2026

    AWS revenue continues to grow as cloud demand remains high

    5 February 2026
  • Apps

    After backlash, Adobe reverses shutdown of Adobe Animate and puts app in ‘maintenance mode’

    7 February 2026

    EU says TikTok must disable ‘addictive’ features like infinite scrolling, fix recommendation engine

    7 February 2026

    Here’s how Roblox’s age controls work

    6 February 2026

    Meta is testing a standalone app for its AI-generated ‘Vibes’ videos

    6 February 2026

    Reddit sees AI search as the next big opportunity

    5 February 2026
  • Crypto

    Hackers stole over $2.7 billion in crypto in 2025, data shows

    23 December 2025

    New report examines how David Sachs may benefit from Trump administration role

    1 December 2025

    Why Benchmark Made a Rare Crypto Bet on Trading App Fomo, with $17M Series A

    6 November 2025

    Solana co-founder Anatoly Yakovenko is a big fan of agentic coding

    30 October 2025

    MoviePass opens Mogul fantasy league game to the public

    29 October 2025
  • Fintech

    Stripe Alumni Raise €30M Series A for Duna, Backed by Stripe and Adyen Executives

    5 February 2026

    Fintech CEO and Forbes 30 Under 30 alum indicted for alleged fraud

    3 February 2026

    How Sequoia-backed Ethos went public while rivals lagged behind

    30 January 2026

    5 days left for TechCrunch Disrupt 2026 +1 pass with 50%

    26 January 2026

    50% off +1 ends | TechCrunch

    23 January 2026
  • Hardware

    Kindle Scribe Colorsoft is an expensive but beautiful color e-ink tablet with AI features

    6 February 2026

    Ring brings “Search Party” feature for finding lost dogs to non-Ring camera owners

    2 February 2026

    India offers zero taxes till 2047 to attract global AI workloads

    1 February 2026

    Microsoft won’t stop buying AI chips from Nvidia, AMD even after its own is released, says Nadella

    30 January 2026

    The iPhone just had its best quarter ever

    30 January 2026
  • Media & Entertainment

    From Svedka to Anthropic, Brands Are Making Bold Plays With AI in Super Bowl Ads

    7 February 2026

    “Industry” Season 4 captures tech fraud better than any show on TV right now

    7 February 2026

    Spotify’s new feature lets you explore the story behind the song you’re listening to

    6 February 2026

    The Washington Post retreats from Silicon Valley when it matters most

    6 February 2026

    Spotify is in the business of selling books and adding new audiobook features

    5 February 2026
  • Security

    Senator, who has repeatedly warned of secret US government surveillance, raises new alarm over ‘CIA activities’

    7 February 2026

    Substack confirms that the data breach affects users’ email addresses and phone numbers

    6 February 2026

    One of Europe’s biggest universities was offline for days after the cyber attack

    6 February 2026

    Cyber ​​tech giant Conduent’s hot air balloon data breach affects millions more Americans

    5 February 2026

    Hackers Release Personal Information Stolen During Harvard, UPenn Data Breach

    5 February 2026
  • Startups

    Gradient’s heat pumps get new smarts to enable retrofitting of old buildings

    8 February 2026

    Accel doubles down on Fibr AI as agents turn static websites into one-to-one experiences

    7 February 2026

    ElevenLabs Raises $500M From Sequoia At $11B Valuation

    7 February 2026

    Fundamental raises $255 million in Series A with a new approach to big data analytics

    6 February 2026

    a16z VC wants founders to stop stressing about crazy ARR numbers

    6 February 2026
  • Transportation

    Prince Andrew’s adviser suggested Jeffrey Epstein invest in EV startups like Lucid Motors

    7 February 2026

    Apeiron Labs Takes $9.5M to Flood Oceans with Autonomous Underwater Robots

    5 February 2026

    Uber appoints new CFO as its AV plans accelerate

    5 February 2026

    Skyryse lands another $300 million to make flying, even helicopters, simple and safe

    4 February 2026

    China is leading the fight against hidden car door handles

    3 February 2026
  • Venture

    Peak XV Says Internal Disagreement Has Led to Partner Exits as AI Doubles

    8 February 2026

    SNAK Venture Partners raises $50 million in capital to support vertical acquisitions

    7 February 2026

    Reddit says it’s looking for more acquisitions in adtech and elsewhere

    7 February 2026

    Secondary sales are shifting from founders’ windfalls to employee retention tools

    6 February 2026

    Sapiom Raises $15M to Help AI Agents Buy Their Own Tech Tools

    6 February 2026
  • Recommended Essentials
TechTost
You are at:Home»AI»The EU says incoming rules for general purpose AI can evolve over time
AI

The EU says incoming rules for general purpose AI can evolve over time

techtost.comBy techtost.com11 December 202308 Mins Read
Share Facebook Twitter Pinterest LinkedIn Tumblr Email
The Eu Says Incoming Rules For General Purpose Ai Can
Share
Facebook Twitter LinkedIn Pinterest Email

The political agreement reached by European Union lawmakers late Friday on what the bloc bills as the world’s first comprehensive law to regulate artificial intelligence includes powers for the Commission to adapt the EU-wide AI rulebook to align with developments in the cutting edge sector. has confirmed.

Lawmakers’ choice of term to regulate the most powerful models behind the current boom in AI production tools — which EU law refers to as “general purpose” AI models and systems — rather than using industry-choice terms , such as “fundamental” or “frontier models” — was also chosen with a view to future protection of incoming law, according to the Commission, with the co-legislators preferring a general term to avoid a classification that could be linked to the use of a particular technology (eg machine learning based on transformers).

“In the future, we may have different technical approaches. And so we were looking for a more general term,” a Commission official suggested today. “Foundation models, of course, are part of general purpose AI models. These are models that can be used for a very wide variety of tasks, they can also be integrated into systems. To give you a concrete example, the general-purpose AI model will be GPT-4, and the general-purpose AI system will be ChatGPT — where GPT-4 is embedded in ChatGPT.”

As we reported earlier, the deal agreed by the bloc’s co-legislators includes a low-risk tier and a high-risk tier for regulating so-called general-purpose AI (GPAI) — such as models behind the virus outbreak in AI production tools such as ChatGPT by OpenAI. The trigger for the application of high-risk rules to artificial intelligence production technologies is determined by an initial threshold set in legislation.

Also, as we reported on Thursday, the agreed draft of the EU AI law refers to the amount of computation used to train the models, also known as floating point operations (or FLOPs) — setting the bar for a GPAI to be considered has “high impact capabilities”. at 10^25 FLOPs.

However, during a technical briefing with reporters today to review the political agreement, the Commission confirmed that this is merely an “initial cap”, confirming that it will have powers to update the cap over time through implementing/delegating acts (eg secondary law). He also said the idea is for the FLOPs limit to be combined, over time, with “other benchmarks” to be developed by a new expert oversight body set up within the Commission, called the AI ​​Office.

Why was 25 FLOPs chosen as the high-risk threshold for GPAIs? The Commission suggests that the number was chosen with the intention of capturing current generational frontier models. But he claimed lawmakers did not discuss or even consider whether it would apply to any models currently in play, such as OpenAI’s GPT-4 or Google’s Gemini, during marathon tripartite talks to agree its final shape rulebook.

A Commission official added that, in any case, it will be up to GPAI manufacturers to assess for themselves whether their models meet the FLOP threshold and thus whether they fall under the rules for GPAIs “with systemic risk” or not.

“There are no official sources that say ChatGPT or Gemini or Chinese models are at this FLOP level,” the official said during the press briefing. “Based on the information we have, and with that 10^25 we chose, we chose a number that could really capture, a little bit, the boundary models that we have. Whether it’s a record of GPT-4 or Gemini or others, we’re not here now to claim – because also, in our context, it’s the companies that should come and evaluate the amount of FLOPs or computing power themselves that they have used. But, of course, if you read the scientific literature, many will indicate that these numbers are the most advanced models at the moment. We will see what the companies will evaluate because they are the best to do that evaluation.”

“The rules have not been written with some companies in mind,” they added. “They’re really written with the idea of ​​setting the limit — which, by the way, can change because we have the ability to authorize ourselves to change that limit based on technological development. It could go up, it could go down, and we could also develop other benchmarks that in the future will be the most appropriate for benchmarking the different moments.”

GPAIs that fall into the AI ​​Act’s high-risk tier will face upfront regulatory requirements to assess and mitigate systemic risks — meaning they must proactively test model outputs to shrink real-world (or “ reasonably foreseeable”) adverse effects on public health. security, public safety, fundamental rights or for society as a whole.

While “low-level” GPAIs will face only lighter transparency requirements, including obligations to apply watermarking to AI output.

The watermarking requirement for GPAIs falls under an article that was in the Commission’s original version of the risk-based framework presented in April 2021, which focused on transparency requirements for technologies such as AI chatbots and deepfakes — but now also applies generally for general purpose artificial intelligence systems.

“There is an obligation to try to watermark [generative AI-produced] text based on the latest state-of-the-art technology available,” the Commission official said, detailing the agreed watermarking obligations. “Currently, technologies are much better at watermarking video and audio than text. But what we’re asking for is that this watermarking is done based on state-of-the-art technology — and then we expect, of course, that over time the technology will mature and be like [good] as much as possible.”

GPAI model makers must also commit to respecting EU copyright rules, including compliance with an existing machine-readable text and data mining exemption contained in the EU Copyright Directive — and a mapping of transparency requirements of the Open Source GPAI Act not to extend the exemption from copyright obligations, with the Commission confirming that the Copyright Directive will continue to apply to open source GPAI.

Regarding the Office of Artificial Intelligence, which will play a key role in setting risk classification thresholds for GPAIs, the Commission confirmed that no budget or staffing numbers for the expert body have yet been set. (Though, in the wee hours of Saturday, the bloc’s Internal Market Commissioner, Thierry Breton, suggested the EU was ready to welcome “many” new colleagues as it tooled up this general-purpose AI watchdog.)

Asked about resources for the AI ​​Office, a Commission official said it would be decided in the future by the EU executive taking “an appropriate and formal decision”. “The idea is that we can create a dedicated budget line for the Office and that we will also be able to recruit national experts from the Member States if we wish, on top of contract staff and on top of permanent staff. And some of these staff will also be deployed to the European Commission,” they added.

The Office of AI will work with a new scientific advisory group that the law also creates to help the agency better understand the potential of advanced artificial intelligence models to regulate systemic risk. “We have identified an important role for the establishment of a scientific team where the scientific team can effectively help the Office of Artificial Intelligence understand if there are new risks that have not yet been identified,” the official noted. And, for example, you also point out some notices about models not covered by the FLOP limit that for some reasons could actually create significant risks that governments should look at.”

While the EU executive appears keen to ensure the key details of the incoming law are published, despite the fact that there is no final text yet — because work to consolidate what was agreed by co-legislators during the marathon 38-hour talks completed on Friday night is the next task facing the block in the coming weeks — there could still be some devils lurking in that detail. Therefore, it is worth considering the text that will appear, probably in January or February.

Furthermore, while the full regulation won’t be in place for a few years, the EU will push for GPAIs to comply with codes of practice in the meantime — so AI giants will be under pressure to stay as close to tough regulations as possible which come as far down as possible, through the block’s Covenant AI.

The EU AI law itself likely won’t be in full force until sometime in 2026 — given that the final text, once drafted (and translated into member state languages), must be ratified by final votes in parliament and the Council, after which there is a short period before the text of the law is published in the Official Journal of the EU and another before it enters into force.

EU lawmakers also agreed a phased approach to the law’s compliance requirements, with 24 months to be allowed until the high-risk rules for GPAIs are implemented.

The list of strictly prohibited uses of AI will be implemented earlier, just six months after the law goes into effect — which could potentially mean banning some “unacceptable risk” uses of AI, such as social scoring or style Clearview AI Selfie scraping for facial recognition databases will go live in the second half of 2024, provided no last-minute opposition to the regulation emerges within the Council or Parliament. (For the full list of prohibited AI uses, read our previous post.)

eu ai act gpai rules evolve general general purpose a incoming purpose Regulation of artificial intelligence rules time
Share. Facebook Twitter Pinterest LinkedIn Tumblr Email
Previous ArticleBeeper Mini is back in business after Apple tried to shut it down
Next Article Food tech update: VC funding declines again in Q3 amid fewer deals
bhanuprakash.cg
techtost.com
  • Website

Related Posts

New York lawmakers are proposing a three-year freeze on new data centers

7 February 2026

Benchmark raises $225 million in dedicated funds to double Cerebras

7 February 2026

How artificial intelligence is helping to solve the labor issue in treating rare diseases

6 February 2026
Add A Comment

Leave A Reply Cancel Reply

Don't Miss

Gradient’s heat pumps get new smarts to enable retrofitting of old buildings

8 February 2026

Peak XV Says Internal Disagreement Has Led to Partner Exits as AI Doubles

8 February 2026

New York lawmakers are proposing a three-year freeze on new data centers

7 February 2026
Stay In Touch
  • Facebook
  • YouTube
  • TikTok
  • WhatsApp
  • Twitter
  • Instagram
Fintech

Stripe Alumni Raise €30M Series A for Duna, Backed by Stripe and Adyen Executives

5 February 2026

Fintech CEO and Forbes 30 Under 30 alum indicted for alleged fraud

3 February 2026

How Sequoia-backed Ethos went public while rivals lagged behind

30 January 2026
Startups

Gradient’s heat pumps get new smarts to enable retrofitting of old buildings

Accel doubles down on Fibr AI as agents turn static websites into one-to-one experiences

ElevenLabs Raises $500M From Sequoia At $11B Valuation

© 2026 TechTost. All Rights Reserved
  • About Us
  • Contact Us
  • Privacy Policy
  • Terms and Conditions
  • Disclaimer

Type above and press Enter to search. Press Esc to cancel.