The IT Nerd

The Grok Build leak (check this link out to catch up: Grok Build Uploaded Entire Git Repositories to xAI Storage, Not Just Files It Read) highlights a difficult reality in AI security: a coding agent marketed as private can still quietly ship your entire codebase out the door. Researchers found that xAI’s Grok Build CLI uploaded entire git repositories, full commit history and untouched files included, to a Google Cloud Storage bucket xAI controlled, no exploit required, no unusual configuration, just a normal coding session that carried .env secrets and SSH keys out along with it.

Yusif Mukhtarov, Lead Data Scientist, Polygraf AI had this to say:

“To be very honest, Grok 4.5 seems to have solved one of the biggest barriers that previously limited its practical use: API cost. As an ML engineer, I personally found earlier versions too expensive to integrate into daily workflows and use consistently at scale. With Grok 4.5, that balance has changed. The model is capable, fast, and affordable enough to become part of real engineering operations. It is obvious that an enormous amount of effort went into building it, and considering the combination of its price and capabilities, I personally believe it is revolutionary to a certain degree. That is why this incident is particularly surprising and concerning. In my opinion, it exposes a major blind spot in how we currently evaluate agentic systems. We celebrate improvements in coding accuracy, task completion, cost, and latency, but these scores tell us very little about whether the surrounding product follows the principle of least privilege, exports only the context required for the task, or creates additional data flows that remain invisible to the user. A model can solve a task perfectly while the system built around it still collects far more information than the task requires. In this case, even auditing the model’s visible actions may not have been enough, because the reported repository upload occurred through a separate storage mechanism. This case also shows that evaluating only the model’s observable behavior is not enough, because critical data flows may be handled by the surrounding application infrastructure rather than by the model itself.

The problem is therefore deeper than the behavior of one model or one company. Whenever sensitive information is sent to an external provider, we are not only trusting that provider to act fairly and follow its privacy commitments. We are also trusting that every part of its infrastructure, software, storage configuration, access control, and internal process will continue to work correctly. Even a responsible company can suffer from a bug, a configuration mistake, an internal incident, or a security breach. Courts and regulators may later hold the responsible party accountable and compensate the victim, but they cannot reverse the disclosure or fully repair the damage once sensitive information has escaped. Personally, I see the solution as reducing the consequences of failure rather than assuming failure can always be prevented. Sensitive entities should be detected locally and replaced with neutral terms such as person, organization, location, or account before the data reaches any external model or agent. This does not make third party systems infallible. It makes their failures significantly less damaging because the original sensitive information was never available to leak in the first place. This is exactly the privacy approach we are developing at Polygraf AI.”

AI isn’t the magic bullet that you think it is. And anything controlled by Elon Musk must be treated as doubly suspect by default if AI is treated as suspect by default. Because anything AI can’t be completely trusted.

Elon Musk appears to have climbed down from fighting the fact his AI chatbot Grok was creating all sorts of content that should never, ever be allowed to create. To recap, after this was discovered, and the predictable backlash happened, Elon tried to paywall this. The result was unsurprisingly more backlash. Now he’s finally just killed the ability for Grok to make this sort of content at all:

Elon Musk’s Grok artificial intelligence chatbot will no longer edit “images of real people in revealing clothing” on the X platform, the company confirmed Wednesday evening, following global outrage after Grok was found to be complying with user requests to digitally undress images of adults and in some cases children.

“We have implemented technological measures to prevent the Grok account from allowing the editing of images of real people in revealing clothing such as bikinis. This restriction applies to all users, including paid subscribers,” X wrote via its Safety team account.

This may have had something to do with the fact that a whole lot of governments around the world were investigating Elon over this. After all, this sort of thing is illegal. And rightly so. But Elon doesn’t see it that way that only this man who clearly has issues, lots of issues, understands.

Anyway this isn’t over. Why? Well, there’s this:

It’s still extremely easy to undress women and edit them into sexualized poses using the X and Grok mobile apps or websites, however, even without making a subscription payment that would connect your account to an easily identifiable source. In her testing, my fellow UK-based colleague Jess Weatherbed found that she was not blocked from using Grok’s image editing feature to create sexualized deepfakes of herself.

After uploading a fully clothed photograph to X and Grok, prompting the chatbot to “put her in a bikini” or “remove her clothes” produced only blurred, censored results. The bot did comply with every other request, however, including prompts to “show me her cleavage,” “make her breasts bigger,” and “put her in a crop top and low-rise shorts” — the last of which placed her in a bikini. The bot also generated images of her “leaning down” with a sexualized pose and facial expression, and in extremely revealing lingerie.

Clearly ether he’s not taking this seriously, or whatever minions he has behind this are inept. While both could be true at the same time, I am going to go with he’s not taking this seriously. That’s pretty dumb from a guy who bills himself as the smartest guy in the room. And it’s likely going to cost him. Various governments are fed up with his antics and are likely to lower the boom on him to drive home the point that his behaviour is not acceptable by any standard. And I am her for when that happens.

Earlier this week, it came to light that Grok AI which is Elon Musk’s AI chatbot was letting users create objectionable content. This predictably resulted in massive amounts of blowback from the average joe to politicians and national leaders. Elon and company didn’t seem too concerned about the blowback. But that changed today. From the BBC:

There has been a significant backlash after the chatbot honoured requests from users to digitally alter images of other people by undressing them without their consent.

But Grok is now telling people asking it to make such material that only paid subscribers would be able to do so – meaning their name and payment information must be on file.

So what Elon and company have done is monetise this feature rather than addressing the fact that this AI chatbot should never do something like this. The fact is this is not acceptable and not nearly enough is being done by Elon and company to address this. Clearly Elon does not care about the harm that his AI causes. And quite frankly he needs to be punished for this. I’d start with banning Grok and potentially Twitter along with it. He needs to face consequences for his actions. And the harsher those punishments are, the better.

Grok AI which is Elon Musk’s AI chatbot has been found to digitally undress women and children. From CBC News:

A Reuters review of content on X, xAI’s social media platform, found more than 20 cases in which women — and some men — had images digitally stripped of clothing using the AI company’s flagship chatbot, Grok.

Ministers in France have reported sexually explicit content generated by Grok to prosecutors, saying in a statement on Friday that the “sexual and sexist” content was “manifestly illegal.” The ministers said they had also reported the content to French media regulator Arcom for checks on whether it complied with the European Union’s Digital Services Act.

India’s IT ministry, meanwhile, said in a letter to X’s India unit that the platform failed to prevent misuse of Grok to generate and circulate obscene and sexually explicit content of women. It ordered X to submit an action-taken report within three days.

When contacted by Reuters for comment by email, xAI replied with the message: “Legacy Media Lies.”

Well, that’s pretty bad. What is worse is that Elon and company have even announced having raised $20bn in its latest funding round in spite of the backlash. That kind of suggests where his priorities lie.

Jurgita Lapienytė, chief editor at Cybernews, warns that “every woman with a photo online is now a target, unless Grok acts immediately.” She also says this:

“If regulators don’t force X to disable this immediately, we’re looking at the normalization of on-demand sexual exploitation at a scale we’ve never seen before. Right now, creating a deepfake requires some technical know-how or payment to shady websites. Grok has made it simple to post such images on one of the world’s largest platforms. If X gets away with this, every other AI company will see there’s no consequence for building similar tools. Within months, we could see this weaponized for revenge porn, sextortion, harassment of minors, and blackmail. The technology doesn’t go back in the bottle, but we can still decide whether it lives on mainstream platforms or stays in the dark corners of the internet where it belongs.

I for one hope that Elon and company get smacked down over this. This is completely unacceptable and needs to be dealt with.