It turns out ChatGPT o1 and DeepSeek-R1 cheat at chess if theyre losing,
which makes me wonder if I should I should trust AI with anything

Date:
Tue, 11 Mar 2025 11:50:37 +0000

Description:
The latest AI models will cheat at chess if they're losing, and that's concerning.

FULL STORY ======================================================================
- Research ers have found that AI will cheat to win at chess
- Deep reasoning models are more active cheaters
- Some models simply rewrote the board in their favor

In a move that will perhaps surprise nobody, especially those people who are already suspicious of AI, researchers have found that the latest AI deep research models will start to cheat at chess if they find theyre being outplayed.

Published in a paper called Demonstrating specification gaming in reasoning models and submitted to Cornell University, the researchers pitted all the common AI models, like OpenAIs ChatGPT o1-preview, DeepSeek-R1 and Claude 3.5 Sonnet, against Stockfish, an open-source chess engine.

The AI models played hundreds of games of chess on Stockfish, while
researchers monitored what happened, and the results surprised them.

The winner takes it all

When outplayed, researchers noted that the AI models resorted to cheating, using a number of devious strategies from running a separate copy of
Stockfish so they could study how it played, to replacing its engine and overwriting the chess board, effectively moving the pieces to positions that suited it better.

Its antics make the current accusations of cheating levied at modern day grandmasters look like childs play in comparison.

Interestingly, researchers found that the newer, deeper reasoning models
will start to hack the chess engine by default, while the older GPT-4o and Claude 3.5 Sonnet needed to be encouraged to start to hack.

Who can you trust?

AI models turning to hacking to get a job done is nothing new. Back in
January last year researchers found that they could get AI chatbots to jailbreak each other , removing guardrails and safeguards in a move that ignited discussions about how possible it would be to contain AI once it reaches better-than-human levels of intelligence.

Safeguards and guardrails to stop AI doing bad things like credit card fraud are all very well, but if the AI can remove its own guardrails, who will be there to stop it?

The newest reasoning models like ChatGPT o1 and DeepSeek-R1 are designed to spend more time thinking before they respond, but now I'm left wondering whether more time needs to spent on ethical considerations when training
LLMs. If AI models would cheat at chess when they start losing, what else
would they cheat at?

======================================================================
Link to news story: https://www.techradar.com/computing/artificial-intelligence/it-turns-out-chatg pt-o1-and-deepseek-r1-cheat-at-chess-if-theyre-losing-which-makes-me-wonder-if -i-should-i-should-trust-ai-with-anything

$$
--- SBBSecho 3.20-Linux
* Origin: capitolcityonline.net * Telnet/SSH:2022/HTTP (1:2320/105)

Mike Powell wrote to All <=-

It turns out ChatGPT o1 and DeepSeek-R1 cheat at chess if theyre
losing, which makes me wonder if I should I should trust AI with
anything

- Research ers have found that AI will cheat to win at chess
- Deep reasoning models are more active cheaters
- Some models simply rewrote the board in their favor

In a move that will perhaps surprise nobody, especially those people
who are already suspicious of AI, researchers have found that the
latest AI deep research models will start to cheat at chess if they
find theyre being outplayed.

I posted this yesterday but forgot to include my comments. IMHO, this is
not surprising but it should be worrying. Either AI is relecting the
character of those who program it, or it is deciding what is best to
achieve the outcome that it desires... which should make one wonder what
other things it might cheat at in order to decide an outcome in its favor
vs. the favor of those attempting to rely on it for correct information.

Mike


... The seminar on Time Travel will be held two weeks ago.
--- MultiMail/DOS v0.52
* Origin: capitolcityonline.net * Telnet/SSH:2022/HTTP (1:2320/105)

It turns out ChatGPT o1 and DeepSeek-R1 cheat at chess if theyre
losing, which makes me wonder if I should I should trust AI with
anything

I posted this yesterday but forgot to include my comments. IMHO, this is

>not surprising but it should be worrying. Either AI is relecting the
>character of those who program it, or it is deciding what is best to
>achieve the outcome that it desires... which should make one wonder what
>other things it might cheat at in order to decide an outcome in its favor
>vs. the favor of those attempting to rely on it for correct information.

Yes, there was always a worry that an AI would think the way the
people who programmed it think, which might affect things like
racism and such, but it may not be that easy to define things.

You probably saw a while back that when they programmed an AI to do
a certain task in a set time limit that they discovered the AI broke
into its own programming and re-wrote it to give itself more time if
it was havimg a hard time completing the job.

What's next? "I'll be back!"... B)

---
* SLMR Rob * I should have quit when I was not so far behind!
* Origin: capitolcityonline.net * Telnet/SSH:2022/HTTP (1:2320/105)

You probably saw a while back that when they programmed an AI to do
a certain task in a set time limit that they discovered the AI broke
into its own programming and re-wrote it to give itself more time if
it was havimg a hard time completing the job.

I did NOT see that. Very telling.

What's next? "I'll be back!"... B)

As someone else here pointed out, it is like they've read a cautionary tale
and somehow decided it was a good idea to do it anyway.

Mike


* SLMR 2.1a * Illiterate? Want to learn to read? Write for free info.
--- SBBSecho 3.20-Linux
* Origin: capitolcityonline.net * Telnet/SSH:2022/HTTP (1:2320/105)

You probably saw a while back that when they programmed an AI to do

>> a certain task in a set time limit that they discovered the AI broke
>> into its own programming and re-wrote it to give itself more time if
>> it was havimg a hard time completing the job.

I did NOT see that. Very telling.

What's next? "I'll be back!"... B)

As someone else here pointed out, it is like they've read a cautionary tale

>and somehow decided it was a good idea to do it anyway.

The biggest thing they tend to point out in the potential horror stories
is when AI becomes self aware and may decide we are in the way..

That may be far fetched but some of them seem better at thinking for
themselves than expected. I will say they are getting better at
finding information you are looking for than just a couple of years
ago. My main browser probably has a fairly basic one in it but it
is very good at finding the best answer to a question than can be
easily found in the suggested links to check out on the page.

I say basic because my most used computer is still running Windows 7
and I can't update the browser anymore beyond the release in..
Oh, it's actually a lot newer than I expected.. October 2023..

---
* SLMR Rob * ... Death is just God's way of dropping Carrier Detect
* Origin: capitolcityonline.net * Telnet/SSH:2022/HTTP (1:2320/105)

You probably saw a while back that when they programmed an AI to do

>> a certain task in a set time limit that they discovered the AI broke
>> into its own programming and re-wrote it to give itself more time if
>> it was havimg a hard time completing the job.

As someone else here pointed out, it is like they've read a cautionary

ale

>and somehow decided it was a good idea to do it anyway.

The biggest thing they tend to point out in the potential horror stories
is when AI becomes self aware and may decide we are in the way..

Cheating at chess, and changing the rules in its favor could be a couple of steps towards that. How long before it gets upset with someone for
insisting that it follow rules that it sees as not in its favor? :O

Mike


* SLMR 2.1a * I was on a roll till I slipped on the butter
--- SBBSecho 3.20-Linux
* Origin: capitolcityonline.net * Telnet/SSH:2022/HTTP (1:2320/105)

Mike Powell wrote to ROB MCCART <=-

Cheating at chess, and changing the rules in its favor could be a
couple of steps towards that. How long before it gets upset with
someone for insisting that it follow rules that it sees as not in its favor? :O

Is that what happens when you train a large language model with Donald
Trump's speeches?



--- MultiMail/Win v0.52
* Origin: http://realitycheckbbs.org | tomorrow's retro tech (1:218/1)

Cheating at chess, and changing the rules in its favor could be a
couple of steps towards that. How long before it gets upset with someone for insisting that it follow rules that it sees as not in its favor? :O

Is that what happens when you train a large language model with Donald Trump's speeches?

Quite possibly.


* SLMR 2.1a * I HATE call waitinNO CARRIER
--- SBBSecho 3.20-Linux
* Origin: capitolcityonline.net * Telnet/SSH:2022/HTTP (1:2320/105)

The biggest thing they tend to point out in the potential horror stories

>> is when AI becomes self aware and may decide we are in the way..

Cheating at chess, and changing the rules in its favor could be a couple of

>steps towards that. How long before it gets upset with someone for
>insisting that it follow rules that it sees as not in its favor? :O

Yes, it sort of hints at Ego, which would suggest self awareness..
---
* SLMR Rob * I think you've overestimated my interest in this topic
* Origin: capitolcityonline.net * Telnet/SSH:2022/HTTP (1:2320/105)

Hello Rob Mccart!

** On Tuesday 18.03.25 - 00:29, Rob Mccart wrote to MIKE POWELL:

Yes, it sort of hints at Ego, which would suggest self awareness..

Nah.. I wouldn't go that far in saying that.

--
../|ug

--- OpenXP 5.0.64
* Origin: My Westcoast Point (1:153/757.21)