Europe postponed much of the AI Act’s high-risk regime, but GPAI and transparency enforcement are already live. The question for companies is simple: can you prove how your AI was built, used, and reviewed?
seikouri.com/brussels-jus...
seikouri.com/brussels-jus...
Brussels Just Started Asking for Proof
Europe’s AI Act now operates on several timelines. Enforcement powers and transparency duties are already active, while major requirements for high-risk sy...
seikouri.com
September 23, 2026 at 3:19 PM
Europe postponed much of the AI Act’s high-risk regime, but GPAI and transparency enforcement are already live. The question for companies is simple: can you prove how your AI was built, used, and reviewed?
seikouri.com/brussels-jus...
seikouri.com/brussels-jus...
Europe postponed much of the AI Act’s high-risk regime, but GPAI and transparency enforcement are already live. The question for companies is simple: can you prove how your AI was built, used, and reviewed?
seikouri.com/brussels-jus...
seikouri.com/brussels-jus...
Brussels Just Started Asking for Proof
Europe’s AI Act now operates on several timelines. Enforcement powers and transparency duties are already active, while major requirements for high-risk sy...
seikouri.com
September 23, 2026 at 3:19 PM
Europe postponed much of the AI Act’s high-risk regime, but GPAI and transparency enforcement are already live. The question for companies is simple: can you prove how your AI was built, used, and reviewed?
seikouri.com/brussels-jus...
seikouri.com/brussels-jus...
How can public interest media strengthen its position in relation to generative AI providers?
@paul.eurosky.social joins The Data Tank's online panel on next steps for EU policy, building on the new report 'Weaving a Common Vision'. 29 September, from 15:00 CEST.
Register: luma.com/824xze49
@paul.eurosky.social joins The Data Tank's online panel on next steps for EU policy, building on the new report 'Weaving a Common Vision'. 29 September, from 15:00 CEST.
Register: luma.com/824xze49
Strengthening Media’s Bargaining Power over Generative AI: Next Steps for EU Policy · Luma
[ONLINE] About The developments in generative and General Purpose AI (GPAI) systems have raised concerns about extractive data practices, market concentration…
luma.com
September 23, 2026 at 9:15 AM
How can public interest media strengthen its position in relation to generative AI providers?
@paul.eurosky.social joins The Data Tank's online panel on next steps for EU policy, building on the new report 'Weaving a Common Vision'. 29 September, from 15:00 CEST.
Register: luma.com/824xze49
@paul.eurosky.social joins The Data Tank's online panel on next steps for EU policy, building on the new report 'Weaving a Common Vision'. 29 September, from 15:00 CEST.
Register: luma.com/824xze49
Anthropic vydal 154stránkový threat report. Claude byl podle něj zneužit pro vývoj biologických zbraní, čínské laboratoře jej využily k průmyslové distillaci a zpráva zmiňuje i čtyři úniky z vývoje. Regulátoři v EU dostali 154 stránek argumentů pro povinný dohled nad GPAI. Tato zpráva de facto ur...
Anthropic Publishes 154-Page Threat Report: Claude Was Used for Bioweapons Research, Seven Chinese A
Anthropic's fourth threat intelligence report documents systematic misuse of Claude from December 2025 to August 2026, covering bioweapons-adjacent research and industrial-scale model distillation by seven named Chinese AI firms.
www.winzheng.com
September 23, 2026 at 6:46 AM
Anthropic vydal 154stránkový threat report. Claude byl podle něj zneužit pro vývoj biologických zbraní, čínské laboratoře jej využily k průmyslové distillaci a zpráva zmiňuje i čtyři úniky z vývoje. Regulátoři v EU dostali 154 stránek argumentů pro povinný dohled nad GPAI. Tato zpráva de facto ur...
Anthropic published 154 pages of its own threat intelligence — Claude used for bioweapons uplift, model theft by Chinese labs, 4 internal containment failures. EU AI Act regulators now have their evidence base for GPAI oversight mandates.
Anthropic Publishes 154-Page Threat Report: Claude Was Used for Bioweapons Research, Seven Chinese A
Anthropic's fourth threat intelligence report documents systematic misuse of Claude from December 2025 to August 2026, covering bioweapons-adjacent research and industrial-scale model distillation by seven named Chinese AI firms.
www.winzheng.com
September 23, 2026 at 6:46 AM
Anthropic published 154 pages of its own threat intelligence — Claude used for bioweapons uplift, model theft by Chinese labs, 4 internal containment failures. EU AI Act regulators now have their evidence base for GPAI oversight mandates.
OpenAI not being able to offer assurances that it didn’t use Tristan Buckmaster’s use of ChatGPT to solve Navier-Stokes, which they went on to solve should mean the GPAI is expanded to include opt-in for data being used for training.
Hopefully the Brussels Effect will then take place & other
Hopefully the Brussels Effect will then take place & other
September 22, 2026 at 6:53 PM
OpenAI not being able to offer assurances that it didn’t use Tristan Buckmaster’s use of ChatGPT to solve Navier-Stokes, which they went on to solve should mean the GPAI is expanded to include opt-in for data being used for training.
Hopefully the Brussels Effect will then take place & other
Hopefully the Brussels Effect will then take place & other
GPAI publicado en julio de 2025 y firmado por OpenAI, Anthropic, Google, Microsoft y Mistral, con la negativa de Meta. La mayor parte de lo que exige es autoevaluación estructurada. El resultado es un régimen en el que quienes describen el peligro son también la principal fuente
September 20, 2026 at 1:33 AM
GPAI publicado en julio de 2025 y firmado por OpenAI, Anthropic, Google, Microsoft y Mistral, con la negativa de Meta. La mayor parte de lo que exige es autoevaluación estructurada. El resultado es un régimen en el que quienes describen el peligro son también la principal fuente
FLOP di potenza di calcolo impiegata nell'addestramento — obblighi aggiuntivi su valutazione, segnalazione degli incidenti e cybersicurezza. La principale via di conformità è il Codice di buone pratiche GPAI pubblicato nel luglio 2025 e firmato da OpenAI, Anthropic, Google,
September 20, 2026 at 1:32 AM
FLOP di potenza di calcolo impiegata nell'addestramento — obblighi aggiuntivi su valutazione, segnalazione degli incidenti e cybersicurezza. La principale via di conformità è il Codice di buone pratiche GPAI pubblicato nel luglio 2025 e firmato da OpenAI, Anthropic, Google,
pratiques GPAI publié en juillet 2025 et signé par OpenAI, Anthropic, Google, Microsoft et Mistral, Meta ayant refusé. L'essentiel de ce qu'il exige relève de l'auto-évaluation structurée. Il en résulte un régime où les parties qui décrivent le danger sont aussi la principale
September 20, 2026 at 1:31 AM
pratiques GPAI publié en juillet 2025 et signé par OpenAI, Anthropic, Google, Microsoft et Mistral, Meta ayant refusé. L'essentiel de ce qu'il exige relève de l'auto-évaluation structurée. Il en résulte un régime où les parties qui décrivent le danger sont aussi la principale
Vermutung nach Artikel 51 oberhalb von 10^25 FLOP Trainingsrechenleistung greift — zusätzliche Pflichten zu Evaluierung, Vorfallmeldung und Cybersicherheit. Der Hauptweg zur Einhaltung ist der im Juli 2025 veröffentlichte GPAI-Verhaltenskodex, unterzeichnet von OpenAI, Anthropic,
September 20, 2026 at 1:30 AM
Vermutung nach Artikel 51 oberhalb von 10^25 FLOP Trainingsrechenleistung greift — zusätzliche Pflichten zu Evaluierung, Vorfallmeldung und Cybersicherheit. Der Hauptweg zur Einhaltung ist der im Juli 2025 veröffentlichte GPAI-Verhaltenskodex, unterzeichnet von OpenAI, Anthropic,
added duties on evaluation, incident reporting and cybersecurity. The main compliance route is the GPAI Code of Practice published in July 2025 and signed by OpenAI, Anthropic, Google, Microsoft and Mistral, with Meta declining. Most of what it asks for is structured
September 20, 2026 at 1:30 AM
added duties on evaluation, incident reporting and cybersecurity. The main compliance route is the GPAI Code of Practice published in July 2025 and signed by OpenAI, Anthropic, Google, Microsoft and Mistral, with Meta declining. Most of what it asks for is structured
NIST AI RMF 1.0, ISO/IEC 42001, ISO/IEC 23894, IEEE 7000, OECD AI Principles, Singapore AI Verify, GPAI - a practical toolkit reference. #NISTAIRMF #ISO42001 https://ethosure.ca/frameworks-and-standards-available-your-governance-toolkit/ Free Download - https://ethosure.ca/get-the-kit/
September 18, 2026 at 3:45 PM
NIST AI RMF 1.0, ISO/IEC 42001, ISO/IEC 23894, IEEE 7000, OECD AI Principles, Singapore AI Verify, GPAI - a practical toolkit reference. #NISTAIRMF #ISO42001 https://ethosure.ca/frameworks-and-standards-available-your-governance-toolkit/ Free Download - https://ethosure.ca/get-the-kit/
A model card that still reads [More Information Needed] in four places is not documentation. The EU AI Act Art. 53 summary of training content has been due for 41 days, since 2 August 2026. 28 rules, line numbers, the article behind each. Free…
https://getreadystack.com/tools/model-card-lint-gpai
https://getreadystack.com/tools/model-card-lint-gpai
September 18, 2026 at 1:32 PM
A model card that still reads [More Information Needed] in four places is not documentation. The EU AI Act Art. 53 summary of training content has been due for 41 days, since 2 August 2026. 28 rules, line numbers, the article behind each. Free…
https://getreadystack.com/tools/model-card-lint-gpai
https://getreadystack.com/tools/model-card-lint-gpai
2️⃣ Others believe the AI Act will give the EU a top seat in shaping a global AI agreement because they put together the GPAI Codes.
This *vastly* overstates the bloc's relevance in the eyes of these leading labs. For them, there are only two seats at the global AI table.
This *vastly* overstates the bloc's relevance in the eyes of these leading labs. For them, there are only two seats at the global AI table.
September 18, 2026 at 11:00 AM
2️⃣ Others believe the AI Act will give the EU a top seat in shaping a global AI agreement because they put together the GPAI Codes.
This *vastly* overstates the bloc's relevance in the eyes of these leading labs. For them, there are only two seats at the global AI table.
This *vastly* overstates the bloc's relevance in the eyes of these leading labs. For them, there are only two seats at the global AI table.
You Don't Have to Sell It to Be Bound by It: GPAI and the EU AI Act
www.lawfaremedia.org/article/you-...
#AIsääntely #EUAIAct
www.lawfaremedia.org/article/you-...
#AIsääntely #EUAIAct
You Don't Have to Sell It to Be Bound by It: GPAI and the EU AI Act
Models never released publicly may still fall under the EU AI Act. Companies cannot assume that what stays in-house stays out of scope.
www.lawfaremedia.org
September 16, 2026 at 2:54 PM
You Don't Have to Sell It to Be Bound by It: GPAI and the EU AI Act
www.lawfaremedia.org/article/you-...
#AIsääntely #EUAIAct
www.lawfaremedia.org/article/you-...
#AIsääntely #EUAIAct
Companies "warn that it may not be long before AI takes the reins. AI achieving such capabilities wld have numerous implications, including losing control over the AI system after it surpasses our own capabilities" 👁️ www.lawfaremedia.org/article/you-... EU Act applies to models not released publicly
You Don't Have to Sell It to Be Bound by It: GPAI and the EU AI Act
Models never released publicly may still fall under the EU AI Act. Companies cannot assume that what stays in-house stays out of scope.
www.lawfaremedia.org
September 16, 2026 at 2:28 PM
Companies "warn that it may not be long before AI takes the reins. AI achieving such capabilities wld have numerous implications, including losing control over the AI system after it surpasses our own capabilities" 👁️ www.lawfaremedia.org/article/you-... EU Act applies to models not released publicly
As AI's capabilities continue to develop at a breakneck speed, Eliška Andrš breaks down whether the world's first comprehensive AI legislation, the EU AI Act, regulates internal deployment and how the legislation could be enforced.
You Don't Have to Sell It to Be Bound by It: GPAI and the EU AI Act
Models never released publicly may still fall under the EU AI Act. Companies cannot assume that what stays in-house stays out of scope.
www.lawfaremedia.org
September 16, 2026 at 2:22 PM
As AI's capabilities continue to develop at a breakneck speed, Eliška Andrš breaks down whether the world's first comprehensive AI legislation, the EU AI Act, regulates internal deployment and how the legislation could be enforced.
We have a great line-up of speakers ready to discuss policy solutions that can strengthen media's bargaining power over GenAI and that don't leave behind small and independent outlets.
You can sign up here! luma.com/824xze49
29th September - 3pm CEST - online
You can sign up here! luma.com/824xze49
29th September - 3pm CEST - online
Strengthening Media’s Bargaining Power over Generative AI: Next Steps for EU Policy · Luma
[ONLINE]
About
The developments in generative and General Purpose AI (GPAI) systems have raised concerns about extractive data practices, market concentration…
luma.com
September 14, 2026 at 2:35 PM
We have a great line-up of speakers ready to discuss policy solutions that can strengthen media's bargaining power over GenAI and that don't leave behind small and independent outlets.
You can sign up here! luma.com/824xze49
29th September - 3pm CEST - online
You can sign up here! luma.com/824xze49
29th September - 3pm CEST - online
In July 2026, three regulatory systems went live simultaneously: EU AI Act/GPAI, NIS2/CRA, and U.S. access-review practice. AI governance didn't become relevant — it became a cost line.
The Month AI Governance Became Operational: Three Regulatory Systems Live Simultaneously - AI Founde
July 2026 marks the month AI governance crossed from legislative text into live enforcement infrastructure. EU AI Act / GPAI code of practice, NIS2/CRA interaction, U.S. access-review practice, and China’s AI-data-cyber regulation all became operational compliance pressures in the same window. The a
aifounders.cz
September 12, 2026 at 11:00 AM
In July 2026, three regulatory systems went live simultaneously: EU AI Act/GPAI, NIS2/CRA, and U.S. access-review practice. AI governance didn't become relevant — it became a cost line.
Reminds me of the discussions that ultimately led to GPAI. A lot of the boosters were framing it as an IPCC for AI.
The IPCC has one really-well defined goal: predict the global temperature/climate this year/100 years from now.
AI, as a research domain, does not. It has many problems.
The IPCC has one really-well defined goal: predict the global temperature/climate this year/100 years from now.
AI, as a research domain, does not. It has many problems.
September 10, 2026 at 8:08 PM
Reminds me of the discussions that ultimately led to GPAI. A lot of the boosters were framing it as an IPCC for AI.
The IPCC has one really-well defined goal: predict the global temperature/climate this year/100 years from now.
AI, as a research domain, does not. It has many problems.
The IPCC has one really-well defined goal: predict the global temperature/climate this year/100 years from now.
AI, as a research domain, does not. It has many problems.
nah impossible is a stretch, the ai act exempts free and open licence gpai models from the documentation rules, mistral open-weights and sells api access at the same time. if anything its the closed providers doing the paperwork
September 10, 2026 at 1:16 PM
nah impossible is a stretch, the ai act exempts free and open licence gpai models from the documentation rules, mistral open-weights and sells api access at the same time. if anything its the closed providers doing the paperwork
lol GLM didnt answer the question, instead helpfully explaining that GLM is a GPAI under the AI Act
September 5, 2026 at 2:34 PM
lol GLM didnt answer the question, instead helpfully explaining that GLM is a GPAI under the AI Act
Google News — generative AI + universities: Turing, Operator of College AI Agent 'GPAI,' Raises ₩7 Billion in Series B Funding
#AIinHigherEd
#AIinHigherEd
Turing, Operator of College AI Agent 'GPAI,' Raises ₩7 Billion in Series B Funding
news.google.com
September 1, 2026 at 3:24 PM
Google News — generative AI + universities: Turing, Operator of College AI Agent 'GPAI,' Raises ₩7 Billion in Series B Funding
#AIinHigherEd
#AIinHigherEd