'Dangerous': Hate-fueled activist raises alarm as Meta sets him loose on AI
Meta’s announcement earlier this month that anti-trans activist Robby Starbuck “will work collaboratively” with the company to address bias in its AI products marks another step in the social media giant’s rapid shift to the right.
Starbuck is a former music video editor who repositioned himself as a conservative influencer, best known for leveraging social media to pressure companies such as Tractor Supply Co. to abandon commitments to diversity, equity and inclusion.
Starbuck has also spread anti-LGBTQ messaging, equating trans people with pedophiles through repeated use of the term “groomer.”
“Robby Starbuck pushes a dangerous anti-LGBTQ+ agenda, spreading disinformation and denying the very existence of transgender people,” Eric Bloem, Human Rights Campaign’s vice president for workplace equality, told Raw Story.
“There’s nothing unbiased about that. Coupled with its January rollback of protections against hate speech across its platforms, this decision calls into question Meta’s commitment to keeping LGBTQ+ people and others safe online.”
Starbuck gained a seat at Meta’s table by suing the company, which owns Facebook, Instagram, Threads and WhatsApp, over false claims by its AI chatbot that he was involved in the Jan. 6 2021 riot at the U.S. Capitol.
Starbuck said in an Aug. 8 post on X that after he filed a defamation suit, “Meta reached out to me immediately, which led to many very long calls with concerned executives and engineers.”
Starbuck and Meta said in a joint statement the same day that “since engaging on these important issues with Robby, Meta has made tremendous strides to improve the accuracy of Meta AI and mitigate ideological and political bias.”
The statement also said “Meta and Robby Starbuck will work collaboratively in the coming months to find ways to address issues of ideological and political bias.”
Starbuck described the settlement “as a win for everyone,” adding that it “produces a better product for Meta” and also “allows me to deliver on multiple fronts as a voice for conservatives.”
But in a statement to Raw Story, he insisted that while he’s made no secret of his political views, he’s not out to impose his beliefs on Meta’s users.
“That would be antithetical to my beliefs about AI, which are that it’s here to stay and needs to show no bias, not my bias, not your bias, not anyone’s bias,” he said. “It needs to be a neutral, fact-driven system.”
‘I hope this is a joke’
Over the past four years, Starbuck has made a string of posts on X labeling LGBTQ people, particularly trans people and people involved in drag performances as “groomers.”
One 2023 post attacked KitchenAid’s sponsorship of trans TikTok influencer Dylan Mulvaney, saying: “KitchenAid will forever be GroomerAid in my house from this day forward.”
In another post, Starbuck called Lil Nas X, whose real name is Montero Lamar Hill, “a groomer and a predator” in response to the rapper’s 2021 video simulating a lap dance with Satan.
“I don’t hate gay people,” Starbuck posted in May 2024. “I hate behaviors that hurt kids. I want people to stop pushing LGBTQ propaganda on kids and stop transitioning kids.”
Starbuck has also openly embraced the Great Replacement theory, a set of racist talking points on immigration closely associated with white supremacist agitation and mass shootings.
Brenton Tarrant, who livestreamed a slaughter of 51 Muslims at two mosques in Christchurch, New Zealand in 2019, named his manifesto “The Great Replacement.”
In February 2024, Starbuck wrote on X: “You can’t call replacement theory racist when it’s literally out in the open now.
“I’m Latino and I’m telling you that the west is trying to replace existing citizens (mostly white) with migrants from 3rd world countries. It must end or the west will become third world!”
Asked about that post in the context of his new role helping Meta guard against bias in AI products, Starbuck told Raw Story: “I hope this is a joke because I’m Latino.
“Trying to associate me to white supremacy or mass shooters is as sick as it is devoid of intelligence.”
A Meta spokesperson declined to comment, other than to reference the joint statement previously issued with Starbuck.
Alejandra Carballo, a clinical instructor at Harvard Law School’s Cyberlaw Clinic, told Raw Story that Meta engaging Starbuck in “any advisory capacity” was “pretty egregious.”
“It’s so incredibly far from where Meta was a few years ago, where Meta was holding stakeholder meetings with LGBTQ groups,” Carballo said.
“It fits in with their tack to the right since the election. They view anti-LGBTQ content as something they’re not only able to tolerate, but something they’re actively greenlighting.”
In January, less than two weeks before Donald Trump’s inauguration, Meta rolled out changes to eliminate third-party fact-checking and weaken policies against hate speech.
Meta’s new policy on Hateful Conduct carved out an exception for LGBTQ people, allowing allegations of mental illness, in contrast to other groups with protected characteristics.
The policy also lifted a prohibition against the anti-trans slur “t----y.”
‘Anti-trans sources’
Among 7,000 Meta users in 86 countries surveyed by the LGBTQ advocacy group GLAAD, along with Ultra Violet and All Out, 72 percent reported that harmful content targeting protected groups has increased since Meta relaxed regulation of hate speech.
Ninety two percent said they felt less protected from being exposed to, or targeted by, harmful content, and 77 percent said they felt less safe expressing themselves freely.
Caraballo said Meta’s Llama chatbot stands out among its competitors “for incorporating far more anti-trans sources.”
Noting that Facebook, Meta’s predecessor, was accused of amplifying hate against the Rohingya people in Myanmar, culminating in a 2017 massacre, Caraballo said she worries that WhatsApp, a platform owned by Meta and popular in the global South, could magnify hate and instigate violence against trans people.
“I can imagine someone like Starbuck being brought in and saying trans people don’t even qualify as a group or people or they’re mentally ill,” Caraballo said.
“The implicit bias in the Llama model could be made even worse.”
At the same time, Caraballo said she saw Meta’s arrangement with Starbuck as more a function of gauging the political winds than pursuing a political agenda.
“Maximizing engagement and minimizing political liability” is the social media giant’s ultimate aim, Caraballo said.
That fits with the decision by Meta in April 2024 to hire Dustin Carmack, chief of staff to the director of national intelligence in the first Trump administration, as director of public policy for the Southern and Southeastern U.S.
Carmack, who was also a senior advisor for the presidential campaign of Florida Gov. Ron DeSantis, authored a chapter of Project 2025, a policy blueprint for the second Trump administration.
In his contribution to the 900-page document, Carmack accused some CIA employees of “promoting divisive ideological or cultural agendas,” and said the new CIA director — who turned out to be John Ratcliffe, his old boss as Director of National Intelligence — “should direct resources from any activities that promote unnecessary and distracting social engineering.”
In July, Meta promoted Carmack to a new job in Washington: director of public policy for the executive branch.
