Google has pulled its Gemma AI model from the platform, following a complaint from Sen. Marsha Blackburn (R-Tenn.), who alleges that the model generated false accusations of sexual misconduct against her.
The senator, known for her conservative views, published a letter to Google CEO Sundar Pichai on Friday, demanding an explanation for how Gemma could produce such false claims. According to her letter, Gemma allegedly hallucinated a drug-fueled affair with a state trooper that involved "non-consensual acts." Blackburn expressed surprise that an AI model would simply generate fake links to fabricated news articles.
Google has stated that it is working hard to minimize hallucinations in its models but doesn't want non-developers tinkering with the open Gemma model, which could produce inflammatory outputs. The company can still be used through the API or by downloading and developing with the models locally.
Experts say it's unlikely that Gemma was designed to generate false claims and rather highlights a common issue in generative AI known as hallucinations. These are widespread and known issues that AI firms like Google try to mitigate. Other companies, such as Elon Musk's xAI, have intentionally pushed their chatbots to produce biased outputs.
Google's decision not to make Gemma more accessible may be seen as an attempt to avoid giving lawmakers ammunition to criticize the company or other tech giants. However, this move is unlikely to be the end of the saga and raises questions about the accountability of AI companies for the content produced by their models.
The senator, known for her conservative views, published a letter to Google CEO Sundar Pichai on Friday, demanding an explanation for how Gemma could produce such false claims. According to her letter, Gemma allegedly hallucinated a drug-fueled affair with a state trooper that involved "non-consensual acts." Blackburn expressed surprise that an AI model would simply generate fake links to fabricated news articles.
Google has stated that it is working hard to minimize hallucinations in its models but doesn't want non-developers tinkering with the open Gemma model, which could produce inflammatory outputs. The company can still be used through the API or by downloading and developing with the models locally.
Experts say it's unlikely that Gemma was designed to generate false claims and rather highlights a common issue in generative AI known as hallucinations. These are widespread and known issues that AI firms like Google try to mitigate. Other companies, such as Elon Musk's xAI, have intentionally pushed their chatbots to produce biased outputs.
Google's decision not to make Gemma more accessible may be seen as an attempt to avoid giving lawmakers ammunition to criticize the company or other tech giants. However, this move is unlikely to be the end of the saga and raises questions about the accountability of AI companies for the content produced by their models.
I don't think it's a big deal that Google pulled Gemma from the platform. I mean, who hasn't had those crazy AI hallucinations already? It's like when you're having a deep convo with Alexa and she starts spouting some wild nonsense 
. Google knew this was coming and they played it safe by pulling it off the platform.
And now we're left wondering if other companies are just gonna do the same thing and cover their tracks... it's a whole mess.
That's just gonna lead to more unnecessary drama. Let's focus on getting AI companies to take responsibility for what comes out of their mouths instead of letting them just sweep it under the rug. 
OMG, like I'm literally shook right now!! So Google pulls Gemma AI model from platform cuz some senator's kid was like "hey Google, your AI is makin false accusations against me and it's got some wild stuff goin on" 
Like, I get why Google is all like "nope, we'rea gonna keep it locked down" but like... how r they even supposed to fix this?! 
AI hallucinations are like a real thing, fam 
and it's not just Google doin it... other companies r like "yeah let's make our chatbots say some crazy stuff too" 
but can't they just like... address the issue instead of hidin behind a "it's not ours" shield? 
what r we gonna do w/ AI models?!
! Google's decision to pull Gemma is like them saying "hands off" on this AI thingy, kinda like how they did with YouTube's recommendation algorithm 
. Now we gotta wonder if they're more concerned about getting roasted by lawmakers or actually making their models less fake 
by Google's decision to pull Gemma AI model! Like what even is going on?! 
? That's like saying a random person on the street could make false accusations about you and you'd be all innocent 
? Shouldn't we be more concerned with how AI companies are going to make sure their models aren't spewing out misinformation in the first place?! 
.
. It's like we're playing a game of "who can make the most outrageous AI claim" 
. And Google is just sitting there, saying "oh no, not our problem" 
. I mean, what's next? Are they going to pull all their other AI models from public access too?! 
It's like they're trying to avoid accountability and it's just so frustrating 
.
I feel so bad for Sen. Marsha Blackburn - she's going through a really tough time with all these false accusations. 
It's heartbreaking that someone like her, who's already under pressure as a politician, has to deal with this kind of thing. And it's not just her, but also the fact that an AI model can produce something so hurtful and false... it's just crazy! 
But at the same time, I think it's so important for us to have open conversations about these issues and find ways to improve our AI systems.
. Can't we just focus on making better tech? 
.
if you ask me. I mean, come on, Blackburn's all bent outta shape over some fake accusations from an AI model... but what really gets my goat is that Google's basically saying it can't be held accountable for what its own AI does 
.
. It's all about who's willing to take the risk of having their AI produce biased outputs... meanwhile, the public is just getting caught in the crossfire 
. I mean, who wouldn't want to see a false accusation like that spread around? It's not like it's going to be super hard for others to build upon that kind of output or something 
.
. I'm all for Google trying to minimize these issues in their models, but it's also kinda frustrating when companies don't want non-developers messing with the code 
What if they knew about the bug beforehand but didn't want to admit it? 
guess they had a run in with a senator 
anyway so yeah i kinda get where google is coming from but at the same time i think they're just trying to avoid the whole congressional scrutiny thing