Diokno’s AI Governance Act Adds a Second Data Center Gate

What It Means

• The AI Governance Act creates the Philippines’ first comprehensive AI law, filed by Akbayan Rep. Chel Diokno alongside Reps. Perci Cendaña, Dadah Ismula, and Kaka Bag-ao.
• Employers deploying high-impact AI must give workers six months’ notice and cannot rely on AI alone for evaluation, discipline, or termination.
• AI data center operators face a new Environmental Compliance Certificate requirement, community consultation, and mandatory public reporting of energy and water use, jointly enforced by DICT, DENR, and DOE.
• Violations carry six months to six years in prison, with permanent disqualification from public office for non-compliant officials.
• The bill’s real structural weight sits in the data center provisions, not the labor rules, since it lands directly on the AI infrastructure buildout already exposed to grid capacity risk.

---

Diokno filed House Bill 10362 at the start of the 20th Congress’s second regular session, reviving an effort to give the Philippines a unified AI Governance Act. The bill’s labor protections have drawn most of the early coverage. The bigger story sits elsewhere.

The AI Governance Act Has Two Different Targets

Most of the AI Governance Act reads like a labor bill. Employers deploying high-impact AI systems must notify affected workers and their representatives at least six months before rollout, spelling out which job functions change, how work assignments shift, and what redeployment or reskilling support will be offered. AI cannot be the sole basis for performance evaluation, discipline, or termination. Employers cannot use AI to sidestep security of tenure, minimum wage, or the right to organize. A new Artificial Intelligence Multi-Stakeholder Advisory Council, drawing from government, academia, labor, and the tech sector, gets a seat in shaping how the rules get written from here.

None of that is enforceable yet, and Philippine labor legislation has a long track record of arriving with strong text and weak implementing rules. The Adjustment Measures Program under DOLE Department Order 241-24 shows the pattern: a program built around worker protection that ended up funding the equipment it was designed to guard against, because the funding categories were never defined narrowly enough. A six-month notice mandate sounds firm until it meets an agency writing the rules that decide what counts as “high-impact,” and that definitional fight is where most Philippine bills of this kind lose their teeth.

The second half of the bill works differently. AI data center and facility operators must secure environmental compliance certification, consult with local communities, and publicly report annual energy and water consumption, standards jointly set by DICT, DENR, and DOE. That is not aspirational language. It is a licensing gate, layered on top of standard industrial ECC requirements that already apply to large facilities.

The Gate Lands on a Buildout Already Under Pressure

Philippine AI infrastructure investment is not hypothetical. The New Clark City corridor is already drawing multi-gigawatt data center commitments, the kind of capital that requires years of planning against known grid capacity and Meralco rate exposure. That exposure was already priced in. Investors backing this buildout knew they were negotiating power costs and distribution capacity years out.

What they had not priced in was a second, independent compliance gate specific to AI facilities, arriving on a separate legislative track from the energy and rate questions already working through DOE and the ERC. If HB 10362 advances, a data center operator now clears grid capacity negotiations, an ERC rate environment in flux, and a distinct AI-specific ECC and disclosure regime, each with its own timeline and its own risk of delay.

This does not kill the buildout. It adds a second regulatory dependency to projects that already carry a long one. Capital that had modeled a single point of regulatory friction now has to model two, and the second one did not exist six months ago. The joint standard-setting arrangement between DICT, DENR, and DOE also means three agencies now have to agree on what counts as compliant before any operator can move, a coordination requirement that has slowed simpler industrial permitting in the Philippines before.

The bill does not specify a phase-in period for existing or already-permitted facilities, which leaves open whether projects already in construction would need to retrofit consultation and disclosure processes after the fact. That ambiguity is itself a cost. Capital markets price uncertainty, not just confirmed obligations, and an undefined transition window sits on the balance sheet the same way a confirmed one would, just with a wider range of outcomes.

The New Exposure Lands on Two Different Actors

Two groups face costs that did not exist before this filing. AI infrastructure developers and their capital partners in the New Clark City corridor now carry environmental compliance and community consultation risk on top of grid and rate risk, with no clarity yet on how DICT, DENR, and DOE will coordinate the joint standard the bill calls for.

Mid-sized BPO and call center operators using AI for scheduling, quality monitoring, or performance scoring face a different exposure. Many have no documented human-review layer sitting behind their AI tools, because none was previously required. If the bill clears committee on anything close to its current timeline, that documentation gap becomes a compressed retrofit problem rather than a planning exercise.

The labor provisions will get the headlines because they are the easiest to explain. The data center provisions are the ones that change what it costs to build AI infrastructure in the Philippines, and that cost lands on capital already committed before the bill existed.

Track more regulatory shifts that affect your business in Policy & Regulation section of Hemos PH.