Unique: Southeast Asia eyes hands-off AI guidelines, defying EU ambitions

SINGAPORE/STOCKHOLM, Oct 11 (Reuters) – Southeast Asian nations are taking a business-friendly method to synthetic intelligence regulation in a setback to the European Union’s push for globally harmonised guidelines that align with its personal stringent framework.

Reuters reviewed a confidential draft of the 10-member Affiliation of Southeast Asian Nations’ (ASEAN) “information to AI ethics and governance,” whose content material has not beforehand been reported.

Three sources advised Reuters the draft is being circulated to know-how corporations for suggestions and is anticipated to be finalised on the finish of January 2024 in the course of the ASEAN Digital Ministers Assembly. Corporations which have obtained it embrace Meta (META.O), IBM (IBM.N), and Google (GOOGL.O).

EU officers earlier this 12 months toured Asian nations in a bid to persuade governments within the area to observe its lead in adopting new AI guidelines for tech companies that embrace disclosure of copyrighted and AI-generated content material.

In distinction to the EU’s AI Act, the ASEAN “AI information” asks corporations to take nations’ cultural variations into consideration and doesn’t prescribe unacceptable threat classes, in keeping with the present model reviewed. Like all ASEAN insurance policies, it’s voluntary and is supposed to information home laws.

With nearly 700 million folks and over a thousand ethnic teams and cultures, Southeast Asian nations have broadly divergent guidelines governing censorship, misinformation, public content material and hate speech that might doubtless have an effect on AI regulation. Thailand, for instance, has legal guidelines in opposition to criticising its monarchy.

Expertise executives say ASEAN’s comparatively hands-off method is extra enterprise pleasant because it limits the compliance burden in a area the place present native legal guidelines are already complicated and permits for extra innovation.

“We’re additionally happy to see this information aligns carefully with different main AI frameworks, corresponding to america’ NIST AI Danger Administration Framework,” IBM Asia’s vice chairman of presidency affairs Stephen Braim stated, referring to voluntary tips developed by the U.S. Division of Commerce’s Nationwide Institute of Requirements and Expertise.

Meta and Google didn’t reply to request for remark.

BENEFITS VS HARM

The information, which is supposed to be periodically reviewed, urges governments to help corporations via analysis and improvement funding and units up an ASEAN digital ministers working group on AI implementation.

Senior officers in three ASEAN nations stated they’re bullish on the potential of AI for Southeast Asia and consider the EU has been too fast to push for regulation earlier than the harms and advantages of the know-how are totally understood.

The ASEAN information advises corporations to place in place an AI threat evaluation construction and AI governance coaching, however leaves specifics to corporations and native regulators.

“We see it as placing ‘guardrails’ for safer AI,” one official advised Reuters. “We nonetheless need innovation.”

The information warns of the dangers of AI getting used for misinformation, “deepfakes”, and impersonation, however leaves it to particular person nations to work out one of the simplest ways to reply.

Different Asian nations corresponding to Japan and South Korea have flagged equally relaxed approaches to AI regulation, casting doubts over the EU’s ambition to determine a worldwide commonplace for AI governance based mostly on the principles that might apply to its 27 member states.

Driving the EU push are issues in Brussels in regards to the fast tempo of AI improvement and its impact on civil rights and safety, which have put threat controls and enforcement on the centre of the proposed laws.

Whereas ASEAN doesn’t have any powers to make legal guidelines, its choice for member states to make their very own coverage determinations places these nations on a distinctly completely different monitor to the EU.

The EU’s struggles to create international consensus on AI regulation distinction with its largely profitable marketing campaign final decade to determine knowledge safety legal guidelines which have develop into a template for different main economies all over the world.

“What we predict is essential is to have comparable rules,” a European Fee spokesperson advised Reuters. “We’re not in search of full harmonisation, as we’re aware of cultural variations, nonetheless, we regard the underlying rules as essential.”

EU officers and lawmakers advised Reuters that the bloc would proceed to carry talks with Southeast Asian states to align over broader rules.

“If we would like AI for use for good, we have to come collectively on the essential rules of human rights,” Dutch minister for digitalisation Alexandra van Huffelen advised Reuters. “I don’t assume we’re very far off from that we couldn’t bridge the variations.”

Our Requirements: The Thomson Reuters Belief Rules.