The EU’s AI Act is now in force
TechCrunchJUL 20 2024
The US, EU, UK, Israel, and six other nations signed the first-ever legally binding international artificial intelligence treaty during a Council of Europe (COE) summit in Vilnius on Thursday.
The document was initially passed in May, having been negotiated between the Council of Europe's 46 members (including all EU countries) and 11 external nations.
The framework's "general obligations" include a commitment to protect human rights, democratic institutions and processes, as well as freedom of speech during AI activities. The text also calls for "adequate" transparency, oversight, accountability, and responsibility.
Having witnessed a series of voluntary agreements in the UK and South Korea, the shift to legally binding commitments is a welcome step forward in the bid to appease growing concerns over the dangers AI may pose to humanity. While tech regulation is not without its critics, and questions remain over ensuring effective implementation, the treaty is nonetheless a landmark milestone.
FollowingHaving pressurewitnessed froma countriesseries suchof asvoluntary agreements in the USUK and South Korea, the COE'sshift publicto sector-focusedlegally conventionbinding commitments is unfortunatelya filledwelcome withstep caveatsforward andin exemptions.the Withbid privateto companiesappease mostlygrowing excludedconcerns andover the usedangers ofAI ambiguousmay languagepose likely to hinderhumanity. enforcementWhile tech regulation is not without its critics, theand AIquestions treatyremain isover farensuring fromeffective whatimplementation, manythe hadtreaty hopedis andnonetheless expecteda landmark milestone.
AIFollowing hysteriapressure hasfrom ledcountries tosuch aas demandthe forUS, regulation far too early in the technologyCOE's development.public Whilesector-focused AI'sconvention capabilitiesis mayunfortunately certainlyfilled onewith daycaveats reachand aexemptions. levelWith ofprivate sophisticationcompanies ormostly dangerexcluded thatand requiresthe intervention,use evidenceof fromambiguous Europelanguage showslikely that,to forhinder nowenforcement, thisthe willAI likelytreaty onlyis stiflefar researchfrom andwhat innovationmany outhad ofhoped fearand of an unlikely existential crisisexpected.
ThereAI ishysteria has led to a 62%demand chancefor thatregulation anfar internationaltoo early in the technology's development. While AI's regulatorycapabilities agencymay forcertainly theone oversightday reach a level of transformativesophistication AIor systemsdanger willthat berequires establishedintervention, beforeevidence 2030from Europe shows that, accordingfor tonow, thethis Metaculuswill predictionlikely communityonly stifle research and innovation out of fear of an unlikely existential crisis.