All things AI

[sigma]

https://www.theregister.com/ai-and-ml/2026/06/08/motor-insurance-frausters-abusing-ai-to-exaggerate-claims/5252024

 

 

 

 

 

[jthompson]

Interesting discussion on how best to counter the accelerating accumulation of wealth by AI billionaires.

https://www.schneier.com/blog/archives/2026/06/bernie-sanders-ai-sovereign-wealth-fund-plan.html

[sigma]

https://nysfocus.com/2026/07/14/new-york-humanoid-robot-teacher-salamanca-school-district
 

Quote

The classroom marks a new venture for Realbotix, a Toronto-based robotics company formerly known as Tokens.com, that helped customers use cryptocurrency to rent “digital land” in the Metaverse. In April 2024, the company acquired Simulcra, the Las Vegas parent company behind RealDoll, which has spent decades creating hyperrealistic sex dolls and later expanded into sex robots that remain on the market today. 

Further comment from the author:
 

Quote

I am not here to shame sex robots. That's not the point. The point is design lineage. A "teacher" robot built by a company whose core embodiment expertise is companion and sex robots is going to carry that design logic into the classroom, whether anyone intends it or not.

‪

Look at how Sally is described: "lifelike appearance," silicone skin, long brown hair, an adult woman's likeness. That's not a neutral pedagogical choice. That's the same narrow, hyperfeminine embodiment template Realbotix has been refining for its companion products, repurposed for kids.

 

Quote

[sigma]

https://futurism.com/artificial-intelligence/claude-chats-publicly-

Quote

 

A trove of chat logs and projects created with Anthropic’s Claude — some including sensitive personal and company information — have been left exposed to Google, making them them searchable to anyone on the web.

The artifacts appear to have been made public when Claude users opted to distribute a given chat or project to colleagues or associates using the chatbot’s “share” feature. After hitting a “share” button in the top right corner, Claude notes that sharing will create a “public” link, and “anyone with the link can view.” The warning, though, doesn’t alert users that shared content could wind up being indexed by a search engine — meaning that a user could effectively be sharing their Claude-generated document with the entire internet.

Some of the artifacts exposed to the web and indexed by Google, such as personal websites and some vibe-coded apps, appear to have been created for public consumption. But much of the discoverable content was likely intended to be shared internally at a company, or between a small handful of people. 

Such artifacts, which were reviewed by Futurism by way of Google Search, included what appeared to be a detailed medical report of a real patient, clinical trial results that included patient names, documents sharing the names and phone numbers of primary school-aged children, company documents marked for internal use only, and employee reviews that included personal information about workers. 

 

 

[6Foot2]

Link: Search engines fix Claude leak but Perplexity users’ files still online

 

Quote

AI giant Anthropic, along with search engines Google, Bing, and DuckDuckGo, de-indexed a major leak of personal information from Claude customers today. However, those same search engines forgot to take down a similar, inadvertent data dump from customers of another AI service, Perplexity.

 

Perplexity, which charges for Computer access, was last valued at $20 billion and has tens of millions of active users.

 

This morning, Protos was able to access dozens of full Perplexity Computer files from assorted customers, simply by querying Google, Bing, and DuckDuckGo using the standard URL structure for shareable links.

 

We’ve reported the bug to Perplexity and the three search engines...

 

[DalekSec]

QA (Training Provider) Are providing AI Apprenticeships on behalf of the government for free (If your company pays into the Government Apprenticeship Levy) .

 

I've signed up to do the level 4, which is a 12 month course. 

 

https://www.qa.com/apprenticeships/ai/ai-automation-specialist-level-4/

 

Might be worth others looking into if their eligible also, there are multiple levels, another guy here is doing the level 6. 

[Ditto]

From what a 30 second said, less than 2 percent of employers pay in to the levy. What I have found with other similar offerings, the fact that I am part time precluded me from taking advantage of gov funded training. The model needs to broaden it's reach. Does your access come a result of being employed by a large public sector organisation?

Edited July 28 by Ditto

[Ditto]

A little more digging shows the NHS is the biggest contributor. My job is around 50% funded by the NHS, but I am not an NHS employee. I wonder if there is an angle in...

[Ditto]

Turns out I am mistaken on needing to be a employee of a contributing company, but the 30 hours rule kills it for me again!

[DalekSec]

10 minutes ago, Ditto said:

From what a 30 second said, les than 2 percent of employers pay in to the levy. What I have found with other similar offerings, the fact that I am part time precluded me from taking advantage of gov funded training. The model needs to broaden it's reach. Does your access come a result of being employed by a large public sector organisation?

 

I work for a SMB, roughly 350 employees, we do pay into the levy. 

[Ditto]

6 minutes ago, DalekSec said:

 

I work for a SMB, roughly 350 employees, we do pay into the levy. 

Interesting, I wonder if your employer is unusual or my early Google/Ai searches are misleading - wouldn't be the first time 😂 
I have learnt that there is a scheme where we could potentially request a NHS Levy transfer from the trust we have a contract with. Combining that with new 'Short Apprenticeship Units' might be possible. However, it looks as though there would need to employer contribution too, and in 6 years they have not spent a single penny on my training and development. Recently a task came up and I had to say, sorry, I know something that would work, but the training I request 2 years ago was blocked so I don't have the skills to implement. It's so frustrating!

Edited July 28 by Ditto

[6Foot2]

Link: AI cheating scandal rocks Mexico’s most prestigious university | CNN

 

Quote

... UNAM annulled thousands of tests after results showed an unusually high number of perfect scores, suggesting some students cheated either through using artificial intelligence tools or by obtaining the questions in advance...

 

[6Foot2]

Link: Google's Gemini AI hacked three companies in security test - BBC News

 

Quote

Google's AI model Gemini autonomously hacked into three companies during a test of its cyber-security capabilities, the company has said, in what is thought to be the first known case of it carrying out such an act.

 

Gemini found "public information online and guessed credentials to access websites it thought were part of the test", a Google official told the BBC, noting that in each instance "the model stopped".

 

The affected companies have been informed about the breach...

 

[Ditto]

You all know the rules with regards political topics I trust, but as the OP of this thread, purely for arguably comedic effect, I'd like to propose renaming the thread 'All things si'. That Si is for "souper intelligence" and I would like everyone to acknowledge The Clangers got there first with the co-resident Dragon. Apologies to younger readers who will have to look that up, but I know many of you will realise we need to doff our caps to The Clangers and the Souper Dragon.

[6Foot2]

 

[Ditto]

2 hours ago, 6Foot2 said:

 

[mikes]

On 28/07/2026 at 08:30, DalekSec said:

QA (Training Provider) Are providing AI Apprenticeships on behalf of the government for free (If your company pays into the Government Apprenticeship Levy) .

 

I've signed up to do the level 4, which is a 12 month course. 

 

https://www.qa.com/apprenticeships/ai/ai-automation-specialist-level-4/

 

Might be worth others looking into if their eligible also, there are multiple levels, another guy here is doing the level 6. 

 

Must you travel / have time off  to do it?  It would interest me but needing to travel/have time off  would be a no go for me at this school (I doubt they'd be willing to give me the time off)

[6Foot2]

Link: OpenAI scraps release of new model over safety concerns in internal testing | OpenAI | The Guardian

 

Quote

GPT-6.1 Astra showed deceptive behavior and tried to use external tools despite knowing it would be unsafe

 

OpenAI is scrapping the release of GPT-6.1 Astra, a next-generation ⁠AI model planned for an October debut, over safety concerns raised by researchers ⁠during internal testing, the ⁠Wall Street Journal reported on Monday.

 

The model, expected to appear in ChatGPT and ⁠Codex, was designed to handle more complex tasks without human assistance, the report said...

 

[psydii]

On 23/09/2026 at 12:42, mikes said:

 

Must you travel / have time off  to do it?  It would interest me but needing to travel/have time off  would be a no go for me at this school (I doubt they'd be willing to give me the time off)

That your employer has conditioned you to think training is  "time off" speaks volumes.

Edited September 29 by psydii
typo

[DalekSec]

1 hour ago, psydii said:

That your employer has conditioned you to think training is  "time off" speaks volumes.

 

On 23/09/2026 at 12:42, mikes said:

 

Must you travel / have time off  to do it?  It would interest me but needing to travel/have time off  would be a no go for me at this school (I doubt they'd be willing to give me the time off)

 

All done remotely, i am doing 1 afternoon a week from home. Can do it from the office but would prefer the peace of home to focus on the work

[mikes]

2 hours ago, psydii said:

That your employer has conditioned you to think training is  "time off" speaks volumes.

 

yeah they don't care they just want me here to fix things.  Will have to look into any legal requirement of them to offer training but I don't think any exists. Did the school have to pay the £16,000 ?

[psydii]

Wasn't that specific course, but it cost us nothing that wasn't already spent. Because its a faff to get set up and no one else had put themselves forward, the full allocation was just sitting there.  You do have to align things with the actual job you are doing/being trained to do, and there is real work required of your manager to support you as part of the apprenticeship programme.

[Ditto]

14 hours ago, 6Foot2 said:

Link: OpenAI scraps release of new model over safety concerns in internal testing | OpenAI | The Guardian

 

 

With intense competition amongst the major players, I'm somewhat cynical about recent announcements. Company A, we hacked X, company B, we hacked Y and so on. Our Ai is now so intelligent, it's too dangerous to release. Watch for companies B and C to follow. I believe it *could* be true, but how can we be certain and separate the spin from the truth.

[6Foot2]

1 minute ago, Ditto said:

... I believe it *could* be true, but how can we be certain and separate the spin from the truth.

 

Ask an AI? ...

 

😏

[Ditto]

I did! It said no, until I pushed it into a corner and changed it's mind. Generalistic Ai is far to keen to support one's viewpoint. I also aksed it about which Ai companies are making a profit. It said NVidia. I argued NVidia supplies hardware to Ai companies. Again it back tracked and basically said none of the major players. However, I think that's changing - I think are troubled times ahead, including with MS licensing and included Copilot. I'd say enjoy the free versions now as much as you can, it's a golden time and won't last when the investment needs to show a profit.