back to article OpenAI warns folks over GPT-4 Vision's limits and flaws

OpenAI is rolling out upgrades for GPT-4 that will, among other things, allow the AI model to answer queries from a user about a submitted image – and the super-lab has at least documented some safety risks involving that ability. The aim of this new functionality is simple: a user can upload a picture file, and via ChatGPT …

  1. Andy The Hat Silver badge

    A glowing indicator give a feeling of control ...

    " customers will always know when Alexa is listening to their request because the blue light indicator will glow"

    Oh, the blue light has gone red ... Alexa, stand down ... STAND DOWN!

  2. Zippy´s Sausage Factory
    Devil

    I know how this will go down

    Developers: "cool new feature but be aware of its limitations"

    Sales: "this amazing new feature is perfect and will solve all your problems, even the ones it can't handle"

  3. tiggity Silver badge

    Mistral

    Don't see a problem with it not being moderated, the censorship on a lot of these AI systems can be very excessive, interesting to see what differences there are with an unmoderated instance if asking anything remotely "controversial"

    .

  4. amanfromMars 1 Silver badge

    Who Dares Win Wins Stealthy Advanced IntelAIgently Resourced Services ‽ .

    Talking of unforeseen or unintended consequences as we are here, Katyanna, with future live incidents which will undoubtedly be constantly redefining of a new field of AI security research committed to intensive research and development of mitigating strategies and solutions destroying attackers risking attacks against AI, is the following page full of pertinent impertinent information and intelligence ........ https://ur2die4.com/uncategorized/231002/ ...... offered for careful consideration as it embraces and presents more than just if it were written as a fiction, ... "CAISER meets Merlin the AIMagician and Mega Meta Data Base Physicians and SMARTR Virtual Clinicians." .... for it features a number of crossroads where AI tools and SMARTR AI-based technologies are verging on the almightily invulnerable and designedly inscrutable.

    Is there any defence against a SMARTR AI adopting existing federal agencies software and capabilities rather than they adopting it?

    cc .... To whom it may concern responsible for leading progress and generating wealth in the UKGBNI re relatively exclusive oversight privileges to readily available, novel remotely controlled future utilities, facilities and abilities.

    That's where IT and AI is currently at, Katyanna, and there be no good reason why nobody should know, given what it is so capable of easily doing practically immediately once it shares what is being done and how it is done and to be done.

  5. NoneSuch Silver badge
    Pint

    My inquiry is being held up.

    Some Just-Stop-Oil activists are sitting on my Internet cable.

POST COMMENT House rules

Not a member of The Register? Create a new account here.

  • Enter your comment

  • Add an icon

Anonymous cowards cannot choose their icon

Other stories you might like