Deepfakes and AI Nudes: What Philippine Law Already Covers

You do not have to wait for a Philippine AI Act. Here is what applies today, and where the real gap is.

Last reviewed: August 26, 2026General legal information, not legal advice
News hook: The 20th Congress has roughly 40 House bills and 7 Senate bills on artificial intelligence, including Senator Pia Cayetano's Senate Bill 25 creating a National AI Commission and Senator Bam Aquino's Senate Bill 758 on deepfakes, which would establish that a person has exclusive control over their image, voice, and identity and give victims a right to compensation. Lawmakers are also working to harmonize House Bills 7396 and 7913 into a Philippine AI Act. As of August 2026 none has been enacted. In the meantime, several existing statutes already reach most deepfake abuse.

Legal question

Someone made a fake nude image or video of you using AI, or cloned your voice, or put your face in a video you were never in. Is that already a crime in the Philippines, and what can you do today?

Applicable laws and rules

Why this matters

The tools are now trivially accessible. Generating a convincing fake nude, a cloned voice sample, or a video of someone saying something they never said takes minutes and no technical skill, and the material spreads faster than any takedown process operates.

The most common reaction — that the Philippines has no law on this because the AI bills have not passed — is wrong, and it costs victims the window in which evidence can still be preserved and accounts still traced. Several statutes apply squarely, and the penalties are serious.

There is nonetheless one real gap, and it is worth naming precisely rather than pretending it does not exist: the voyeurism law was written for real recordings, not synthetic ones.

If the subject is a child, the analysis stops here

RA 11930, the Anti-OSAEC and Anti-CSAEM Act of 2022, is the controlling law whenever the depicted person is a minor, and it is deliberately drafted to reach synthetic material. The definition of child sexual abuse or exploitation material extends to representations of a child in real or simulated explicit sexual activities, and to material produced by digital or any other means. It also treats as a child a person depicted or presented as a child.

The consequences are severe — long imprisonment and heavy fines — and the law imposes duties on internet intermediaries, hosting providers, and financial institutions to detect, block, and report. Possession alone is punishable, and there is no defense that the image was generated rather than photographed. Report to the PNP Women and Children Protection Center, the NBI, or the Inter-Agency Council Against OSAEC.

Adults: the Safe Spaces Act is usually the best fit

For adult victims of sexualized deepfakes, Section 12 of RA 11313, the Safe Spaces Act, is generally the strongest existing hook. It penalizes gender-based online sexual harassment, which includes the uploading or sharing without the victim's consent of any form of media containing photos, voice, or video with sexual content, the unauthorized recording and sharing of a victim's photos, videos, or information online, impersonation, and the use of information and communications technology to terrorize or intimidate through unwanted sexual remarks, threats, or misogynistic slurs.

The advantage of the Safe Spaces Act here is that its focus is on the non-consensual sexualized use of the person's likeness and identity online, rather than on whether a particular recording is authentic. That framing accommodates synthetic material more comfortably than the voyeurism statute does.

Identity theft under the Cybercrime Prevention Act

Section 4(b)(3) of RA 10175 penalizes computer-related identity theft: the intentional acquisition, use, misuse, transfer, possession, alteration, or deletion of identifying information belonging to another, whether natural or juridical, without right.

A person's face and voice are identifying information. Using them to generate content the person never created — a fake video, a cloned voice message, a fabricated endorsement — falls within the terms of the provision. Where the deepfake is used to obtain money, computer-related fraud under Section 4(b)(2) also applies, which covers the increasingly common scam of a cloned voice calling a relative or a finance officer with an urgent transfer request.

Section 6 of the same law raises by one degree the penalty for any crime under the Revised Penal Code or a special law committed through ICT. The Supreme Court's July 16, 2026 ruling on psychological violence committed through Facebook confirms how directly that operates — provided the ICT element is alleged in the Information.

If the perpetrator is a partner or ex-partner

RA 9262 applies where the perpetrator is a husband, former husband, a person with whom the woman has or had a sexual or dating relationship, or the parent of a common child. Creating or circulating sexualized fake images of a former partner is psychological violence causing mental or emotional anguish, public ridicule, or humiliation under Section 5(i).

The practical advantage is speed: RA 9262 carries protection orders. A barangay protection order can be obtained at the barangay and issued the same day; a temporary protection order can be issued by the court ex parte. These can order the respondent to stop the conduct and to stay away, well before any criminal case is resolved.

The gap: RA 9995 was written for real recordings

RA 9995, the Anti-Photo and Video Voyeurism Act, is the law most people reach for, and it is the weakest fit. It prohibits taking a photo or video of a sexual act or of the private areas of a person without consent, and separately prohibits copying, selling, distributing, publishing, or broadcasting such material — with the copying and distribution offenses applying even where the original recording was consensual.

The difficulty is that the statute is built around an actual image of the person's body. A deepfake typically maps a real face onto a body that is not the victim's, or generates the body entirely. On a strict reading, no photo or video of that person's private areas was ever taken. That is a genuine interpretive gap, and it is precisely the gap the pending deepfake bills are meant to close.

Two qualifications. First, where real intimate images exist and were merely altered or recombined, RA 9995 applies directly. Second, the gap in RA 9995 does not mean no law applies — it means the charge should be built on the Safe Spaces Act, identity theft under RA 10175, RA 9262 where the relationship exists, and the civil provisions below.

The civil route, which is often underused

Article 26 of the Civil Code requires every person to respect the dignity, personality, privacy, and peace of mind of others, and expressly makes actionable — for damages, prevention, and other relief — acts such as meddling with or prying into the private life of another, and vexing or humiliating a person. Articles 19, 20, and 21 supply the general framework: a person who willfully causes loss or injury to another in a manner contrary to morals, good customs, or public policy must compensate for the damage.

A civil action for damages under these provisions does not require the conduct to be a crime, applies a lower standard of proof than a criminal case, and can be joined with a request for injunctive relief. Moral and exemplary damages are available. For a victim whose principal harm is reputational and emotional, this route is often faster and better matched to the injury than a criminal complaint.

Separately, the Data Privacy Act treats images and biometric information as personal information. A complaint to the National Privacy Commission can be brought against a person or entity processing your personal data without a lawful basis, and the NPC can order corrective measures.

Election deepfakes

Comelec has already regulated this. Resolution No. 11064, adopted for the 2025 national and local elections and the BARMM parliamentary elections, banned the use of AI and fake social media accounts as campaign material and prohibited AI bots used to spread disinformation for or against a candidate. Where AI is used, the resolution required a clear, conspicuous disclaimer that cannot easily be removed or altered, including a statement that every individual depicted in AI-manipulated material gave prior consent. Violations are election offenses, carrying one to six years imprisonment without probation, disqualification from public office, and loss of the right of suffrage. Comelec's Task Force KKK sa Halalan monitors compliance with the NTC and the CICC.

That resolution governed the 2025 cycle. Anyone campaigning in the November 2, 2026 barangay and SK elections should assume Comelec's position on AI-manipulated material has not softened, and should check the applicable resolution for that contest before publishing any AI-generated campaign content.

What to do right now

  1. Preserve evidence before requesting takedown. Capture full screenshots including the URL, the posting account, the timestamp, and the comment thread; save the file itself and note where you found it. A successful takedown destroys the evidence if you did not capture it first.
  2. Report to the platform using its non-consensual intimate imagery or impersonation channels, which are usually faster than any legal process.
  3. File with the PNP Anti-Cybercrime Group or the NBI Cybercrime Division. They can request preservation of computer data from service providers, which matters because logs expire. Bring your captures and any leads on the account.
  4. Consider a protection order if the perpetrator is a partner or former partner. This is the fastest relief available.
  5. Consider the civil action for damages under Articles 19, 20, 21, and 26, with a request for injunctive relief, especially where you know the identity of the person responsible.
  6. If the subject is a minor, treat it as an OSAEC matter immediately and report to the PNP Women and Children Protection Center or the NBI; do not forward, store, or share the material for any purpose, including to prove the case.

What individuals should know

You are not waiting on Congress. The Safe Spaces Act, the Cybercrime Prevention Act's identity theft provision, RA 9262 where a relationship exists, RA 11930 where a child is involved, and the Civil Code's dignity and privacy provisions all apply today, and the ICT circumstance raises criminal penalties by one degree. Frame the complaint around non-consensual use of your likeness and identity, not around whether the image is genuine — that framing avoids the RA 9995 gap.

Move quickly. Content spreads, accounts are deleted, and service-provider logs are retained only for limited periods, so a preservation request through law enforcement early is worth more than a perfectly drafted complaint filed late.

Finally, watch the legislation but do not rely on it. Senate Bill 25 would create a National AI Commission with an AI Ethics Review Board empowered to register AI systems, audit high-risk deployments, impose fines, suspend licenses, and recommend criminal prosecution; Senate Bill 758 would establish exclusive control over one's image, voice, and identity and a right to compensation for emotional and reputational harm; Senate Bill 852 would create a Philippine Council on Artificial Intelligence. All remain pending, and the harmonized Philippine AI Act is still being assembled from House Bills 7396 and 7913.

Ask PHLaw.AI

Try: "Someone made fake nude photos of me using AI and posted them on Facebook. What laws can I use to file a case, and how do I preserve the evidence?"

Ask about this topic

Sources