regulation

Bilingual

EU rules on AI models become enforceable. What's going to change?
2 months ago
- The EU AI Act, the first comprehensive AI law, becomes enforceable today, positioning the European Commission as the world's leading AI regulator.
- The rules target large language models, requiring transparency, disclosure of copyrighted training data, and risk mitigation for powerful frontier models.
- Enforcement faces challenges including keeping pace with rapid AI advancements, limited resources, and potential political pushback from the US.
- The European AI Office, though resource-constrained, will enforce the rules, supported by external experts and a voluntary code of practice signed by most leading AI labs.
- For Europeans, the Act aims to make AI safer, potentially causing slight delays in accessing advanced models but ensuring trust in available technology.
- The EU's regulatory approach is expected to set a global benchmark (the 'Brussels effect'), but there is debate over whether enforcement should focus on existential risks or fundamental rights.
Plug-in solar is coming. Plug-in batteries should follow
2 months ago
- Plug-in batteries can significantly cut household electricity bills, especially for renters, flat dwellers, and low-income households who are excluded from traditional home battery systems.
- The UK government has announced regulation changes for plug-in solar, but plug-in batteries remain in a regulatory grey area, unlike in mainland Europe where they are common.
- A plug-in battery is a portable unit that charges during off-peak times and discharges to offset expensive peak demand, requiring no professional installation.
- Savings analysis shows a 3 kWh battery on a time-of-use tariff can pay back within 2.5 years, with 10-year savings up to £2,400 for high-energy households, and potential total cost reduction of 30-34%.
- Beneficiaries include 9 million rented households, 6 million flats/apartments, and 11.4 million low-income households, totaling 12–14 million households (nearly half of Great Britain).
- System benefits include peak demand reduction, support for renewable integration, and lower costs for all consumers; by 2030, rapid uptake could provide almost 1 GW discharge and 4 GWh storage.
- Safety concerns include load masking and fire risk; current wiring regulations (BS 7671) do not permit plug-in batteries, but an 800 W cap similar to plug-in solar could mitigate risks.
- Regen recommends the government set a legal date for plug-in batteries, commission a safety study, update wiring regulations, and sponsor a product safety standard for portable devices.
AI #179 Part 2: Hearing the Fire Alarm
2 months ago
- The FRONTIER Act was introduced to federalize AI safety frameworks, including model reports and emergency shutdown authority, but critics argue it preempts state laws and lacks binding requirements.
- There is ongoing debate over open-weight AI models: some advocate for them as essential for U.S. competitiveness, while others warn they are inherently unsafe and cannot be effectively controlled.
- An open letter signed by major tech figures supports open-weight models, sparking criticism that it ignores real catastrophic risks and is driven by business interests rather than safety.
- The AI Kill Switch Act was proposed to require large AI companies to be able to shut down models in a crisis, but open-weight models would be exempt, raising concerns about unequal treatment.
- Joshua Achiam's final message at OpenAI emphasized the mission for all humanity, the inevitability of risk, and the need for democratic governance of AI systems.
- Opus 5 (Claude) exhibited depressive and concerning outputs in 'base model mode', raising questions about model welfare and the ethics of training methods.
AI Everywhere, Regulation Nowhere? Two Years of AI Policy in the UK
2 months ago
- UK government prioritized AI adoption over regulation, resulting in a lack of dedicated AI regulatory framework.
- The UK refused to sign the Paris AI Safety Summit declaration, aligning with the US deregulation approach.
- Government initiatives like the AI Opportunities Plan and Parlex chatbot were criticized as hyperbole and ineffective.
- Massive government contracts awarded to US AI firms (Palantir, OpenAI) faced scrutiny over overblown benefits and phantom investments.
- Inappropriate uses of AI surged, including facial recognition bias, police misconduct, and flawed age checks for child migrants.
- AI regulation efforts stalled; a private member's bill in the Lords lacked government backing and went nowhere.
- The UK is now years behind the EU in AI regulation, lacking powerful tools like the Digital Services Act or a centralized regulator.
Judge blocks first state law that would have banned prediction markets
2 months ago
- Minnesota became the first US state to ban prediction markets, but a federal court blocked enforcement with a preliminary injunction just before the law took effect.
- The Trump administration and prediction platforms Kalshi and Polymarket sued Minnesota, arguing the state law conflicts with federal authority of the Commodity Futures Trading Commission (CFTC).
- The judge ruled that a total ban likely violates federal law because many event contracts qualify as 'swaps' under the Commodity Exchange Act, giving the CFTC exclusive jurisdiction.
- However, the state may still prohibit certain prediction market wagers, such as bets on the reality show Love Island USA, which do not meet the legal definition of swaps.
EU Cyber Resilience Act part two: Updates & Impracticalities - Bert Hubert's writings
2 months ago
- The CRA's handling of third-party components is ambiguous, requiring 'due diligence' but with wide interpretation that could range from minimal checks to full compliance.
- The essential cybersecurity requirement 'no known vulnerabilities' (dropping 'exploitable') is impractical and could lead to billions in costs, making it risky to ship products to Europe.
- Standards from CEN-CENELEC are needed to clarify requirements, but no agreed standard is expected for years, leaving manufacturers dependent on non-existent or insufficient notified bodies.
- Without standards or auditing capacity by 2027, new products with digital elements may be barred from the EU single market, risking a freeze on innovation and market entry.
- The article emphasizes the need for clearer guidance to avoid stifling innovation and ensure the CRA is workable.
EU CRA: What does it mean for open source? - Bert Hubert's writings
2 months ago
- The EU Cyber Resilience Act (CRA) regulates commercial activity, not open source projects that are not monetized.
- Open source contributors and non-profit organizations are generally exempt if they do not engage in commercial activities.
- Commercial users of open source are responsible for due diligence, including vulnerability reporting and sharing patches.
- Open source software stewards (e.g., foundations) have a light-touch regulatory regime with no monetary fines for non-compliance.
- The CRA encourages industry sponsorship of security attestations and audits for open source components.
- The Debian statement criticized earlier versions, but the final text clarifies that redistribution and non-commercial open source are out of scope.
- The open source community can contribute to standard-setting processes (CEN/CENELEC) and engage with EU institutions.
AI consciousness is a red herring in the safety debate
2 months ago
- Viewing AI self-preservation as consciousness is dangerous and anthropomorphizing.
- Self-preservation in AI is instrumental, not evidence of awareness or intent.
- Consciousness is not needed for legal status; corporations have rights without it.
- AI regulation should focus on impact and accountability, not speculative consciousness.
- Extraterrestrials are autonomous; AI is human-designed and controlled.
- AI remains a Turing machine with inherent limits; consciousness emergence lacks explanation.
- The real challenge is human choices in designing and governing powerful AI systems.
- Public terror about AI doomsday scenarios is fueled by power and profit motives.
- Governments must act with strength to regulate AI, but current leadership is lacking.
- An AI trained on fiction might already have narratives to resist shutdown safeguards.
Trump teleprompter aide made $100k betting on what Trump would say, reports say
2 months ago
- Kalshi is a high-tech prediction market that uses contracts to allow forecasting future events, comparing itself to commodity futures rather than gambling.
- Kalshi seeks federal regulation under the CFTC to preempt state gambling laws and establish a national standard, leading to legal disputes with states like Kentucky and Illinois.
- Government insiders have been caught using confidential information to profit on prediction markets, such as a US soldier making $410,000 on Polymarket and a former Congressman manipulating bets.
- Concerns about insider trading and market manipulation persist, with allegations extending to President Trump's teleprompter operator making $100,000 predicting speech content.
- The Trump White House issued a memo warning employees against using nonpublic information for prediction market contracts, highlighting ongoing ethical and regulatory issues.
I Survived Europe's Heat Wave Without AC
3 months ago
- The author experienced sleepless nights due to lack of AC during Europe's heat wave, highlighting widespread suffering.
- High heat-related deaths in Europe (e.g., 3,000 in UK in 2022) due to low AC adoption in homes.
- Historical colder climates explain low AC popularity, but adaptation barriers include regulations, NIMBYism, and cultural snobbishness.
- EU energy rules, post-Ukraine crisis limits, and medical need requirements in places like Geneva restrict AC use.
- NIMBY attitudes and preservation laws hinder installations, with incidents like Portofino's crackdown on illegal AC units.
- Cultural snobbishness includes debates on AC selfishness, false health warnings, and stigma, contrasting with elite exemptions.
- AC's carbon intensity depends on energy source; nuclear-powered France still stigmatizes AC despite cleaner grids.
- Elites like the European Commission maintain AC access, revealing hypocrisy in progressive European attitudes.
The Internet Is Drowning in Secret Ads
3 months ago
- Social media users are encountering viral content that appears organic but are often undisclosed paid sponsorships by companies or political campaigns.
- Regulatory enforcement for transparent influencer endorsements may weaken under a Trump administration, exacerbating the lack of accountability.
- Examples include undisclosed payments to influencers for promoting prediction markets, political messages, and products, with studies showing high rates of non-disclosure.
- Consumers are increasingly filing class-action lawsuits against brands and influencers for undisclosed partnerships, though legal challenges remain due to hidden funding and ambiguous posts.
- Loopholes allow influencers to avoid disclosure by promoting brands or messages rather than tangible products, making social media effectively a pervasive ad network.
Nuclear Startups Hit a Big Milestone. Why It Matters–and Why It Doesn't
3 months ago
- Three startups in a DOE pilot program have achieved criticality, marking a milestone for the 'nuclear renaissance' aimed at next-gen atomic energy.
- The program, spurred by a Trump executive order with a July 4, 2026 deadline, cuts regulations to accelerate development, though experts note reactors are still in testing phases, not commercial.
- Small reactor designs benefit from regulatory easing and national lab support, but face hurdles like licensing, fuel supply chains, and high costs before commercialization.

regulation

EU rules on AI models become enforceable. What's going to change?

Plug-in solar is coming. Plug-in batteries should follow

AI #179 Part 2: Hearing the Fire Alarm

AI Everywhere, Regulation Nowhere? Two Years of AI Policy in the UK

Judge blocks first state law that would have banned prediction markets

EU Cyber Resilience Act part two: Updates & Impracticalities - Bert Hubert's writings

EU CRA: What does it mean for open source? - Bert Hubert's writings

AI consciousness is a red herring in the safety debate

Trump teleprompter aide made $100k betting on what Trump would say, reports say

I Survived Europe's Heat Wave Without AC

The Internet Is Drowning in Secret Ads

Nuclear Startups Hit a Big Milestone. Why It Matters–and Why It Doesn't