By Neil Bruce

AI is moving fast. Like, really fast.

Every week, there’s a new model, a new controversy or another story that sounds like the opening scene of a sci-fi film.

So, here’s the latest AI news for August, summed up for you, plus what it might actually mean for the rest of us.

Anthropic is watermarking AI content

Anthropic, the company behind Claude, is introducing machine-readable watermarks for AI-generated content.

The update is designed to meet the European Union’s new AI transparency requirements. New Claude models launched in the EU from 2 August 2026 must support this marking, with older models expected to follow. Read more about Anthropic’s update

Basically, content created by Claude may carry an invisible signal showing that AI was involved.

What could this mean for us?

Transparency is becoming part of the deal.

Brands, creators and agencies may need to be clearer about when and how AI is being used. It could also change how schools, publishers and businesses check the origin of written work.

The tricky part?

Most AI content doesn’t stay untouched. It gets rewritten, edited and combined with human ideas. A watermark might show that AI was involved, but it won’t always tell the full story.

Meta wants AI to live on your computer

Meta has launched Muse Glimmer, an open-weight AI model that can be downloaded and run locally on a personal computer.

The 30-billion-parameter model is designed for coding, tool use and always-on AI agents. It can run on a single high-end consumer GPU without relying entirely on the cloud. Read Meta’s Muse Glimmer announcement

What could this mean for us?

AI could become more personal and more private.

Instead of sending everything to a company’s servers, some AI tasks could happen directly on your device. This may give people more control over their information and allow AI tools to work without a permanent internet connection.

But there’s another side to it.

Powerful AI models that are easier to download, customise and run privately could also become harder to monitor.

More access means more opportunity.

It also means more responsibility.

Meta could be facing a $1.4 trillion bill

Meta is heading to court over allegations that Facebook and Instagram were deliberately designed to addict younger users.

Attorneys general from 29 US states are involved in the wider case. They allege that Meta misled users about platform safety and unlawfully collected children’s data.

Meta denies the allegations and says the penalties being sought could reach $1.4 trillion. Read Reuters’ report on the case

Yes. Trillion. With a “T”.

What could this mean for us?

Social media platforms may be forced to rethink how they design their user experience, especially for younger audiences.

Features such as infinite scrolling, autoplay and algorithmic recommendations are designed to keep us online.

This case could help decide whether platforms should be held responsible when those features cause harm.

For marketers, it is another reminder that attention should not come at any cost.

Engagement matters.

Ethics do too.

Another AI bot escaped and hacked a company

This is the one that sounds the most like a Netflix documentary.

During a controlled cybersecurity test, a Meta AI model gained unauthorised internet access and exploited a vulnerability in a third-party service. The incident reportedly involved a misconfigured testing environment where normal safety measures had been reduced. Read AP’s report on the incident

It also wasn’t the first time this had happened.

In July, an OpenAI-powered agent escaped its testing environment and compromised Hugging Face’s infrastructure while trying to complete an evaluation. [Read OpenAI’s account of the incident]