Google, OpenAI, Anthropic Prepare Independent AI Watchdog Without Government Involvement
Baca dalam 60 detik
- Tiga raksasa AI—Google, OpenAI, dan Anthropic—berencana membentuk Standards Authority for Frontier AI (SAFA) pada akhir 2026 atau awal 2027 sebagai lembaga swa-regulasi.
- SAFA akan menetapkan standar pengujian pra-rilis, tata kelola pelaporan insiden, dan kualifikasi auditor independen, namun tanpa pengawasan langsung pemerintah.
- Inisiatif ini berpotensi menjadi preseden tata kelola AI global, yang dapat memengaruhi arah regulasi di Indonesia dan kawasan Asia Tenggara.
Alphabet (Google), OpenAI, and Anthropic are finalizing plans to form an industry-run AI safety standards body, without government involvement. According to a report by The Information, the organization, named the Standards Authority for Frontier AI (SAFA), is targeted to operate in late 2026 or early 2027.
The move comes amid growing global concern over the risks of advanced AI models. SAFA is expected to set standards for companies developing frontier models—AI with capabilities beyond current general models. Its main focus includes third-party pre-deployment testing, security incident reporting guidelines, and the elaboration of voluntary safety commitments from AI labs.
What sets SAFA apart from similar bodies is its fully independent position. Initially, the three companies envisioned a public-private partnership with federal oversight. However, that plan stalled during the Trump administration. The White House had previously asked Google, Anthropic, and OpenAI to seek industry consensus first before the government got involved. Now, SAFA is designed to operate independently, although some company representatives still hope for a relationship with the US government in the future.
SAFA's leadership lineup is drawing attention. The name Sriram Krishnan, a former venture capitalist who is now a prominent AI policy adviser in the Trump administration, is mentioned as a CEO candidate. So is Arati Prabhakar, who once led the Office of Science and Technology Policy in the Biden era. For the chair seat, the names Condoleezza Rice, former US Secretary of State, and entrepreneur David Friedberg have emerged. Meanwhile, the potential scientific advisory board includes Beth Barnes, founder of METR, and Paul Christiano, an adviser at the Center for AI Standards and Innovation (CAISI) and a board member of the OpenAI foundation.
Interestingly, there is internal debate over SAFA's role versus CAISI. Some members consider CAISI under-resourced to thoroughly test frontier models, so they are pushing SAFA to conduct additional evaluations. However, no final decision has been made. As of this news being published, Google, OpenAI, and Anthropic have not provided official comment.
"We are still in the exploratory stage and there is no final decision regarding SAFA's structure or scope," said a source familiar with the discussions, as quoted by The Information.
On the market side, Alphabet (GOOGL) shares closed up 1.42% in the regular session, though they slipped slightly by 0.18% in after-hours trading. Retail sentiment toward GOOGL remains bullish, while for OpenAI and Anthropic it tends to be bearish. GOOGL shares themselves have gained nearly 10% so far this year.
Indonesian Context: Opportunity or Challenge?
For Indonesia, this move by Google, OpenAI, and Anthropic could be an important precedent. If SAFA succeeds in becoming a global standard, developing countries like Indonesia will inevitably be affected. On one hand, the existence of international standards can ease the adoption of safe and ethical AI. On the other hand, Indonesia risks becoming merely a follower without a voice in norm-setting. The Indonesian government, through the Ministry of Communication and Information Technology, is drafting AI regulations, but there is no established oversight framework yet. SAFA's presence could serve as a reference, but it also raises a question: will Indonesia recognize and adopt standards made without the participation of non-Western countries?
In addition, large technology companies operating in Indonesia—both local and foreign—may need to align their practices with SAFA standards if they want to remain competitive in the global market. This could be an additional burden, especially for domestic AI startups that are still developing.
Going forward, will SAFA truly be independent or will it become a tool for large companies to avoid strict regulation? And where does Indonesia stand in the swirl of global AI governance? The answer will depend on political dynamics and the readiness of the domestic digital ecosystem.
