Antonym: Money Wins Edition
Invest responsibly, your species may be at risk.
The whiplash drama of the weekend of OpenAI will boil down to who wins: computer scientists concerned about the potential harms of machine super-intelligence or money. Fans of money vs. other things through history will know which side to bet on (the clue is in the specie).
OPENAI COUP: ALIGN IN THE SAND
My instinct is to talk about OpenAI’s firing of its CEO from a business point of view, but the extremely surprising development may not be about business at all. Sam Altman was (and maybe still will be) the business side of things and the business wants to move very fast indeed.
Computer scientists working on artificial intelligence are often deeply concerned by the implications of their work and possibilities of harm from so-called Artificial General Intelligence (AGI), when AI outstrips human intelligence.
At the big tech companies where many of them work concerns about “alignment” are often the cause of tension with management. Anthropic, a rival to OpenAI, has been able to attract talent from other AI companies because of its perceived strong principles to the concept of aligning AI with humans’ wellbeing. In layperson terms: they make a bigger deal out of avoiding a robot apocalypse.
The move against Altman doesn’t make business sense. It makes sense as a power move and the power is with the scientists, for now. I’m conscious this section make be out of date before I’ve proof-read this newsletter, let alone send.
Will OpenAI will part-own your content from December?
The Control-Altman-Delete news was overshadows so many developments in tech this week, even X’s latest downward spiral. So I’m going to focus on some small print that has definitely not got enough attention.
Logging in to ChatGPT I was asked to accept new terms and conditions that would come into effect in December. Like everyone else I don’t usually bother reading these. This time I thought I would get the AI to analyse its own changes and I built a little bit to compare and contrast them…
Up until this December, if you make something in ChatGPT it belongs to you. After new terms and conditions come in, OpenAI, its parent company may have “certain rights” over what you create.
According to analysis of terms and conditions released today by OpenAI in Europe, which will come into effect from you as a user own the inputs (your prompt, any data or documents you upload), but ChatGPT has rights to the output.
I tested this with a hypothetical example: providing a transcript, such as a presentation or podcast, and then developing that into an article and the GPT (bot) said:
The resulting article, which is a combination of your original transcript and ChatGPT's responses, can be considered a joint creation. In such cases, both you and OpenAI could have claims to portions of the content, depending on the nature and extent of each party's contribution.
So the changes, at least in the EU (& UK) case would be that anything created with AI-assistance would be effectively co-owned by OpenAI and you.
These are the terms and conditions previous version and the new EU version, we assume there will be similar changes coming to other regions. And this is the GPT Small Print —Terms and Conditions Bot that we co-own with OpenAI… I think (#notalwayer).
This may be OpenAI tightening its legal position ahead of the launch of its app store, but at the very least it seems to create a grey area around ownership of content just as many are considering AI adoption.
A consequence may be further legitimate reservations by legal teams. Microsoft is indemnifying users of its platforms based on AI against copyright issues, but this based more on the thorny issue of the IP of the creators of the content that OpenAI trained its models on.
We’ll be watching for further legal expert commentary on this area and recommend checking for ownership and use clauses in any new services you or your team use.
For further reading:
As Generative AI Tools Proliferate, Terms and Conditions Are Becoming a Key Differentiator, By Peter Cramer on October 31, 2023. Proskauer New Media and Technology Law Blog.
Listening
For good reporting mixed with business analysis, Pivot’s emergency podcast on the OpenAI board coup is essential listening.
Watching
Unwanted (NowTV/Sky). A contemporary story that – and sometimes looks – like science fiction. A boat packed with refugees fleeing torture and war in Africa catches fire in the Mediterranean. A passing Italian cruise ship, one of the largest in the world, picks up some of the survivors. When they find out they may be returned to Libya they hijack the ship.
Thanks all for this week folks, but frankly not the half of it. Thanks for reading…
Antony