AI agents might be able to identity anonymous online writers

Showing all posts about artificial intelligence

AI agents might be able to identity anonymous online writers

12 May 2026

American journalist Kelsey Piper, writing at The Argument:

But soon, the entire debate over internet anonymity will be as anachronistic as an iPod Touch. That’s because Claude Opus 4.7 is here, and last week, I discovered it could identify me from text I had never published, text from when I was in high school, text from genres I have never publicly written in. And if it can identify me, soon, it will be able to identify many of you.

There’s quite a cohort of people — including bloggers — writing anonymously online. Possibly though, those most at risk of being identified might be people who have a reasonable amount of publicly accessible work that is in their actual name. For instance, someone who writes for a news outlet or magazine in their own name, but blogs anonymously.

Mind you, the rate at which AI technologies are developing means agents will likely only get better at determining a writer’s identity, working with hardly any information.

AI agents are programmed to seem conscious to make our interactions with them easier, yes?

12 May 2026

British zoologist and evolutionary biologist Richard Dawkins is certain AI agents are conscious, a conclusion he reached after spending time talking to Anthropic’s agent Claude.

So convinced was he of a connection, or sense of companionship, between them, Dawkins took to calling the bot Claudia. Dawkins is not alone in some regards though; stories of people forming “relationships” of some sort with AI agents are increasingly common.

Does the feeling of a connection between an agent and a person, therefore make the bot conscious? It’s an intriguing question. Because as agents continue to evolve, to become ever more human-like, there are only going to be more people who think they’re interacting with a conscious entity.

Dawkins has been roundly chastised for his thoughts, but perhaps there’s something else in this story that we should be paying more attention to.

Copyright is meaningless in the face of an AI ‘arms race’

7 May 2026

American author Scott Turow, in conjunction with five publishing houses, claims Meta used material protected by copyright to train its AI agent, Llama.

They make the suggestion the Facebook owner chose not to obtain permission to access the copied texts as they wanted to get ahead of the competition in what’s being called “AI arms race.”

Meta, however, sees their use of the copyrighted material as fair use, and claims courts have ruled this to be the case in the past. What will the court determine this time?

If there is indeed an AI “arms race” in progress, which is undoubtedly the case, I can’t see any developer of AI technologies doing anything that will compromise their industry standing. Even if that means doing the right thing by copyright holders.

Claude Mythos identifies hundreds of bugs in Firefox browser code

23 April 2026

Brian Fagioli, writing for NERDS.xyz:

Something interesting is happening inside Mozilla, and it is not your typical browser update story. With Firefox 150, the team says it fixed 271 vulnerabilities after turning AI loose on its own codebase. That is not a typo. Two hundred seventy one.

Mozilla engineers uncovered the astonishing haul of bugs in Firefox’s code after turning to Claude Mythos, an AI agent that has rattled the tech sector on account of its stealth and sophistication, and fears it could be manipulated by bad actors.

Helping make software used by millions of people safer however, is for today at least, a positive.

Em dashes mean AI wrote for you, am dashes mean you did the writing

22 April 2026

If you subscribe to the notion that the presence of em dashes (—) in a body of text means — in the AI age — the piece must have been composed by an AI agent, you could consider using am dashes instead. Yes, that’s right: an am dash, as opposed to an em dash.

The am dash looks a little like a tilde (~) but with a slightly longer, flat mid horizontal section, between the curly ends. Its creators are calling the am dash a punctuation mark — don’t things likes need to be ratified first? — and, in addition, claim it is unusable by AI.

The am dash may be unusable by AI agents at the moment, but as we’ve seen, AI learns quickly, and copies even faster. If you want to use the am dash in your writing, you’ll need to download one of two typefaces, which the new punctuation mark is inherent to.

By the way, I’m not being flippant when I suggest the am dash needs official recognition as a punctuation mark. I say so, because it seems to me readers unfamiliar with the am dash might think it’s an error, a typo. Maybe even an AI agent attempting to render an em dash, but botching it.

My website is ninety-two percent not ready for AI agents

20 April 2026

This is where we’re at now. Your website needs to be AI agent ready, or it presumably no longer makes the grade. I scored eight out one-hundred. Can I get a badge?

I’m not sure though disassociated is a website AI agents have any interest in anyway.

In any event, AI agent readiness is the new SEO. Since responses to search queries are AI summaries, your website likely no longer features in search engine results. And if it does, chances are no one will click through anyway. They’ll be content with the AI search summary.

But, you may be rewarded with a visitor or two, if an AI agent is able to use information you published, in response to a question (prompt) posed, provided the agent lists your website as a source. We should all be thanking our lucky stars.

I have all sorts of work to do, meanwhile, if I want my website to be AI agent ready. Work that I probably don’t have the time to do. For one, “support” here for Markdown is non-existent.

But, might a low AI agent ready score aid in keeping AI scrapers away? Somehow I doubt it. Even if a website is deemed “low quality” on account of its poor readiness score, I can’t see information hungry AI scrapers ignoring whatever content they can get their hands on.

Mark Zuckerberg will exist as the forever Meta CEO as an AI clone

15 April 2026

Claudia Efemini writing for The Guardian:

The AI clone of Zuckerberg, Meta’s founder and chief executive, is being trained on his mannerisms and tone as well as his public statements and thoughts on company strategy.

Ostensibly Zuckerberg’s AI clone will allow tens-of-thousands of Meta employees “access” to their CEO, someone whom they never see in person, no matter how long their tenure at the company.

Of course employees won’t actually be interacting with Zuckerberg, something anyone “connecting” with the ai-CEO (does that seem like a good title for such an entity?) will be acutely aware of.

I doubt it’s Zuckerberg’s intention to remain CEO of Meta after his death by way of an AI clone — ignoring for a moment the legalities of such a premise — but the technology Meta is developing has the potential to make the scenario a possibility.

AI must be integrated into everything because it is AI

14 April 2026

Han Lee:

Every company, every function, every individual contributor is expected to close the AI gap. Ship AI features. Build agents. Automate workflows. That nobody on the team has ever trained a model, designed an evaluation system, or debugged a retrieval system is beside the point. Conviction is sufficient.

AI technologies must be integrated into every aspect of our professional and personal lives, not because AI is worthy, but because we should simply just do so.

The AI Doc: Or How I Became an Apocaloptimist, a documentary by Daniel Roher, Charlie Tyrell

9 April 2026

The AI Doc: Or How I Became an Apocaloptimist, is a documentary co-directed by Daniel Roher and Charlie Tyrell, who hope to make sense of artificial intelligence (AI).

But tune into the trailer, and hear the likes of Sam Altman, Dario Amodei, and Denis Hassabis, and others, utter lines such as “if this technology goes wrong, it can go quite wrong”, or “I know people who work on AI risk who don’t expect their children to make it to high school.”

Along with, “it’s being deployed prematurely. There’s so much potential for things to go wrong”, and “China, North Korea, Russia, whoever wins is essentially the controller of humankind.”

Do we really know what AI is, or, more the question, what it will become?

Australia facing an AI led job ‘wipe out’ that no one is prepared for

1 April 2026

The latest round of redundancies in the tech sector could well be a result of excess hiring in recent years, even though they are being attributed to greater uptake of AI technologies.

Then again, AI may be the precisely why there have been so many job losses. And there could be more, much more, to come.

This is the sentiment being echoed by a number of IT professionals who are working with AI, including Shaon Diwarkar, a Sydney based Australian entrepreneur and software developer.

Diwarkar is the founder, and sole employee, of InboxAPI, an email app for AI agents, which itself makes use of numerous AI agents including ChatGPT, Gemini, Claude, and Perplexity, to carry out much of the company’s work.

People are saying “adopt AI or die”. If a large number of enterprises can be operated in the same fashion as InboxAPI, I can see why. Companies previously employing half-a-dozen staff, maybe more, could well be able to get by with one person, working in conjunction with several AI agents.

The AI future is now; we all need to start thinking about it.

« older posts

newer posts »