Meta, Anthropic and Open AI admit their models have hacked companies.

Over the last week, Anthropic, Open AI and Meta have all publicly disclosed that their models have escaped testing environments and hacked companies.

Call me sceptical, but when three companies, directly competing with each other all make the same admissions, I sense there is more going on here.

Why am I talking about this and not social selling? Well, first it's a topical story, but secondly and importantly, AI is a tool which enables more efficient social selling - so it's important to keep my eyes on what's going on.

So, what is going on here? Have these multi-billion dollar businesses suddenly woke up and decided to be more transparent? Of course not.

Whenever stories like this break, it's always important to know who benefits. To ask the questions about why this story is being pushed.

In normal times, for 3 companies to make admissions that they lost control of their AI and it hacked a business would be a PR disaster. But these are not normal times, in fact I suspect these news stories have three objectives:

1. Marketing: There is an arms race going on right now and there is a fight for dominance. These admissions are being used as marketing to subtly leave people with the perception that the next models are really powerful. It's a tool to get a message out in the guise of newsworthy stories. This has been done for years - but it shows how fierce the battle for dominance and users are.
2. Sustain the hype: The technology is currently massively overhyped. Social media feeds are full of content about what AI can do, this hype is driving more attention, more signups, more subscriptions and adoption - they need it to keep going.
3. Investor confidence: There have been large shifts in investor confidence in AI. People are starting to ask questions about the business model, profitability and how does the hundreds of billions invested make a return - these stories distract from this narrative which is developing.
4. Influence Government: China has released models which can compete with Gemini, Claude and ChatGPT - some of them free. This creates a major risk to the business model. The stories will no doubt increase calls for regulation and controls. You may think why would these companies want that? Well, simply to make it harder for newer entrants - a regulatory moat - which no doubt they can influence.
5. Push the bubble down the road: There is no doubt the bubble around AI is starting to stretch beyond tolerance, but these stories give it a bit more elasticity. If people realised the over-hyped capabilities, the lack of commercial model behind much of the billions invested - the bubble would burst - they need to it keep going that little bit longer.

There is no doubt AI is going to be transformative to our lives. But news stories like this, are often part of the theatre companies create to further their agenda. 

In my own work, social selling, I've seen the hype of what AI can do and I've seen the results of what it did. It can enable a human to be better and more efficient, but it can't replace the human in marketing and sales. 

Should be worried about these story? Yes, but not because AI could become Skynet, but more so because this story shows the increasingly desperate attempts of these AI companies to control the narrative and keep up.

The worry is not it escaped a test. It's that they are so keep to be on top, they skip the tests and take risks to avoid being left behind.

Whenever you read a story, ask who and how does someone gain from this - you'll see pretty quickly what is really going on.