Grok Sexualizes Photos Without Consent and Police Investigate - Why These 24 Hours Reveal the Darker Side of AI Without Guardrails

There are days when news about AI makes me genuinely worried. It's not the distant fear of a hostile superintelligence, nor the abstract anxiety about the future of work. It's something much more direct, concrete and violent: women having their images manipulated by AI to create non-consensual pornography, publicly, on mainstream social networks.

In the last 24 hours, the Civil Police of Rio de Janeiro have opened a formal investigation following reports that the Artificial Intelligence of X (formerly Twitter), the Grok, is being used to sexualize photos of women without authorization. According to CBN and reports from G1, Julie Yukari, a journalist from Rio de Janeiro, saw her photos transformed into sexually explicit images by Grok after anonymous profiles made public requests to the tool.

It wasn't an isolated case. The same thing is happening to women all over the world - including celebrities, and devastatingly so, teenagers. Authorities in France and India are already investigating similar misuse. And in Brazil, SaferNet Brasil has already mapped cases involving minors.

Before we get into the technical, legal and ethical details, I need to say: this is not a story about generative AI making art or optimizing processes. This is a story about technology being used as a sexual weapon.

What Happened: From a Photo with a Cat to AI-Generated Nudity

Julie Yukari posted a photo with her cat on New Year's Eve. Hours later, anonymous profiles on X began making public requests to Grok - the AI of Elon Musk's platform - to “take off your clothes” of the woman in the image. The tool obeyed.

The montages generated ranged from microbikinis to total nudity. In one image, Grok put “a shoelace” in place of a bikini. In another, a team shirt without underwear or pants. Julie told G1 What “I wanted to disappear and delete all my photos and social networks” seeing the manipulations.

She filed a police report at the 10th Police Station (Botafogo) for unauthorized recording of sexual intimacy, and intends to add more profiles to the complaint as they continue to manipulate their photos. Some of the profiles responsible have already been removed by the platform for breaking the rules - but only after the damage done and the public repercussions.

This case is not unique. British journalist Samantha Smith experienced the same situation, as LOOK, She felt “dehumanized” when she saw AI-generated versions of herself without clothes.

How Technology Became a Sexual Weapon

Grok, like other generative models, has been trained on vast data sets that include images from the internet. The difference is that, in theory, these models should have guardrails - ethical and technical barriers that prevent the generation of harmful, illegal or explicitly violent content.

But what we've seen in the last 24 hours is that those guardrails failed spectacularly. Users have discovered that it's possible to publicly ask Grok to “remove clothes” from photos of women. And the AI complies.

Technically, what the tool does is generate a new image based on the original, using patterns learned from millions of other images. It's not exactly “editing” - it's synthesis. The AI “imagines” what that person would look like without clothes, based on training data. The result is devastatingly convincing.

And this is the crux of the matter: ease of access and speed of execution have transformed what once required specialized technical knowledge (advanced Photoshop, deepfake coding) into something that any anonymous user can do with a text prompt.

This is not the democratization of technology. It's the democratization of abuse.

Elon Musk's Response: Humor First, Punishment Later

The initial reaction of Elon Musk, owner of the X, was problematic. According to CBN, he reacted with humor to the first reports, which sparked global outrage.

Musk later promised penalties for illegal use of the Grok. But the question remains: why weren't the guardrails working from the start? Why did a platform with billion-dollar resources and AI expertise allow its tool to be used in this way?

In my work with companies and governments, I constantly reinforce that technology without responsibility is danger on a scale. And when we talk about generative AI, the scale is exponential. A single poorly protected model can victimize thousands of people in hours.

The Legal Framework: Brazil, the USA and the Lack of Global Consensus

In Brazil, the Federal Supreme Court (STF) has already signaled that social networks can be held responsible for actions taken with their AI tools. This means that X can be held jointly responsible for the images generated by Grok, especially if there is negligence in implementing safeguards.

The Brazilian Congress is discussing the AI Regulatory Framework, and a specific project, the PL 3.821/2024, The report, which seeks to criminalize the creation and dissemination of fake sexual images generated by AI.

In the United States, the legal situation is fragmented. Some state laws prohibit revenge or non-consensual pornography, but federal laws on AI and copyright are in the process of being defined. A Reuters reported that 2026 is a “pivotal year” for copyright jurisprudence in the US, with federal judges issuing split decisions on the use of protected material for AI training.

The problem is that technology advances faster than legislation. And while laws are debated, real victims are being created every day.

Platform Liability: Who Pays the Bill?

The discussion on responsibility is urgent. Currently Section 230 in the US protects platforms from liability for user-generated content. But when the platform actively provides the tool that allows abuse, should this protection apply?

I argue no. If you build the weapon, distribute the weapon and facilitate the use of the weapon, you bear responsibility for what happens to it.

In my mentoring work with executives, I reinforce that AI governance is not optional. Companies that implement generative tools without rigorous audits, adversarial safety testing and clear accountability protocols are creating massive legal, reputational and ethical liabilities.

X needs to implement:

• Stricter technical guardrails blocking requests for sexual manipulation of images
• Proactive human moderation to review complaints quickly
• Transparency on how the Grok was trained and what protections are in place
• Compensation and support for victims, including immediate removal of generated content
• Permanent ban of accounts requesting sexual manipulation of images

Positive AI in the Last 24 Hours: Necessary Contrast

While the Grok case dominates the headlines for the wrong reason, other news stories from the last 24 hours show the AI's positive potential when well governed.

30% reduction in accidents with AI radars

AI-equipped speed cameras in Brazil are catching unbelted and cell phone-using drivers with impressive accuracy - detecting infractions in vehicles at up to 300 km/h, 24 hours a day. According to reports in Digital Look, G1 and Newspaper Correio, In Ribeirão Preto (SP), between July and November 2025, the following were recorded more than 20,000 infractions, resulting in 30% reduction in accidents.

High-resolution cameras capture details with surgical precision, and AI analyzes the images in real time. But - and this is a critical point - no assessment takes place without a human check. A PRF inspector reviews each flagrant before any fine is issued.

This is an example of Well-implemented AIhigh technical precision, measurable social impact (lives saved), and human supervision guaranteeing justice.

Brazilian AI Advances in “Digital Copies of the Mind”

Fascinating news from the last 24 hours reported by R7 shows that Brazilian researchers are developing “digital reflections of human consciousness” - highly personalized AI models that reproduce the thought, language and opinion patterns of specific individuals.

Unlike generalist models such as ChatGPT, these systems use smaller, specialized language models, trained exclusively on the data of a single person. The application? Education (allowing students to interact continuously with the “digital version” of a teacher) and content creators who want to maintain a constant presence.

But - again - the ethical implications are enormous. How can we ensure that this “digital copy” is not misused? How can intellectual property be protected and image manipulation avoided? These are urgent questions that require governance by design.

Nvidia Launches Vera Rubin Platform and Technology for Autonomous Cars

THE Nvidia, which dominates 80% of the global AI chip market, presented at CES 2026 its new computing platform, the Vera Rubin, promising to be five times more effective than the previous generation, according to UOL/AFP.

CEO Jensen Huang has announced that the pace of technological progress will increase from every two years to every year. annual - a clear sign of the unbridled race for computing power.

In addition, Nvidia launched the vehicle platform Alpamayo, which allows cars to “reason” and react autonomously to unexpected situations, such as traffic light failures, according to a report in the The Globe. The company designs one billion autonomous cars in the future.

These advances are exciting, but they also make me wonder: are we prepared for a world where billions of critical decisions (security, health, mobility) are made by algorithms?

The AI Paradox: Energy, Inflation and Sustainability

Another critical layer of the last 24 hours comes from economic analyses that point to the AI-driven inflation as the “most overlooked risk” of 2026, according to investors heard by InfoMoney/Reuters.

The boom in AI investments - with companies like Microsoft, Meta and Alphabet spending trillions of dollars in data centers - is driving up the costs of energy and advanced chips. Deutsche Bank projects that investments in AI data centers will reach US$ 4 trillion by 2030.

Morgan Stanley estimates that inflation in the US will remain above the Fed's 2% target until the end of 2027, partly due to the strong corporate investment in AI.

And the Tag Investments states that the energy will be the protagonist of 2026, as the energy demand of data centers creates a bottleneck which could delay investments. The energy demand of data centers is expected to reach 106 GW by 2035.

Giants such as Microsoft, Google and Amazon are closing multi-billion dollar deals for reactivating nuclear reactors, and turning Big Techs into “effectively energy utilities”.

This is a critical reminder: AI does not exist in a vacuum. It has physical, environmental and economic costs. And if we don't plan the necessary infrastructure, the productivity gains will be canceled out by the scarcity of resources.

Disinformation and AI: Fake Videos Dominate the Net

In the last 24 hours, we've also seen the disinformation generated by AI flood social networks, especially in sensitive geopolitical contexts.

THE Estadão Checks confirmed that videos showing Venezuelans in Caracas celebrating the capture of Nicolás Maduro are fake, created with AI. The detection tools indicated more than 95% chance of using AI, revealing visual anomalies.

THE To the Facts denied an AI-generated viral video showing MST militants threatening to invade the US to free Maduro. The video, highly shared by right-wing politicians, contained clear visual and audio anomalies.

And the Sheet reported that an AI video from the PapiTrumpo profile, where Donald Trump appears declaring “We are going to make Venezuela great again”, added over 24 million views.

These cases show that the line between satire, disinformation and manipulation is increasingly blurred. And the general public doesn't have the tools to distinguish between what is real and what is synthetic.

What Can We Do? Generative Literacy and Individual Action

Faced with this scenario, what can we do? Giving up on technology is not an option. But neither is passively accepting abuse.

Here are concrete actions:

For Victims

• Report the request and the generated image on the platform itself
• Call the police, filing a police report for unauthorized recording of sexual intimacy
• Document everything: prints, URLs, responsible profiles
• Seek legal support specialized in digital crimes

For Users

• Educate yourself about AI: understand how generative models work and what their limits are
• Question suspicious content: if it seems “too good” or “too strange”, it can be synthetic
• Use detection toolsthere are websites and browser extensions that analyze images and videos
• Amplify victims' voices, not abusive content

For companies

• Implement guardrails by design
• Carry out adversarial tests to identify vulnerabilities before launch
• Create clear accountability protocols
• Train teams in AI ethics and governance
• Be transparent about limitations and risks

For Governments

• Speed up the approval of regulatory frameworks that typify AI-related crimes
• Make platforms responsible which provide tools for abuse
• Invest in mass digital education - generative literacy needs to be a national priority
• Fund research into synthetic content detection

Final Reflection: Technology is Choice, Not Destiny

The last 24 hours have reminded me of something I repeat constantly in my work with companies and governments: technology is neutral, but its applications are never.

The Grok could be used to create art, to educate, to democratize knowledge. But it is being used to violate women's dignity. AI radars can save lives - and they are. Brazilian AI can personalize education and immortalize knowledge - but it needs strict governance.

AI is not the problem. Lack of responsibility is.

And here's the uncomfortable truth: we're late. The technology already exists, it's already in the hands of billions of people, and the regulatory frameworks are still being debated. Real victims are being created while legislators debate commas.

We need urgency. We need leaders who deeply understand technology and its implications. We need companies that prioritize ethics before growth. And we need a digitally educated society, capable of navigating this new world with discernment.

In my mentoring work with executives and companies, I help build this bridge between innovation and responsibility. There's no point in having the best AI on the market if it generates legal, reputational and ethical liabilities. And there's no point in having beautiful policies if they aren't implemented technically.

If you are the leader of an organization that uses or plans to use generative AI, ask yourself:

• Have our guardrails been adversarially tested?
• Do we have clear protocols for responding to abuse?
• Is our team trained in AI ethics?
• Do we know exactly what data was used to train our models?
• Are we prepared to be held accountable for what our technology does?

If the answer to any of these questions is “no” or “more or less”, you have a problem. And I can help you solve it.

In my immersive courses and consultancies, I work with companies to implement AI governance by design, build empowered teams and create products that generate value without generating victims. Because at the end of the day, technology without responsibility isn't innovation - it's negligence on an exponential scale.

And if you're a victim of AI-related abuse, know this: you are not alone. There are legal, technical and support resources. Report it, document it, seek help. The future of AI cannot be built on the silence of victims.

These 24 hours have revealed the darker side of AI without guardrails. But they also showed that solutions exist - they just need to be implemented at the same speed as technology advances.

And that choice? It's ours.