Grok Continues to Push Gender Ideology, Despite Complaints

PRESS RELEASE

Robert Thompson: 301-801-0608

Grok Continues to Push Gender Ideology, Despite Complaints

WASHINGTON / July 31, 2025 – Last week President Donald Trump issued an Executive Order on “Preventing Woke AI in the Federal Government.” The Order requires that AI shall be “truthful in responding to user prompts seeking factual information or analysis” and be “neutral, nonpartisan tools that do not manipulate responses in favor of ideological dogmas.” (1)

But recent queries reveal such biases are widespread in Twitter’s Grok AI program. Following are examples of recent exchanges with Grok:

Gender Wage Differences

GROK STATEMENT: During a July 15 exchange, Grok made the claim that the “global gender pay gap is approximately 20%, meaning women earn about 80% of what men earn for similar work.” (2)

FACTS:

Numerous studies show when work experience, hours worked, risks of injury, and other factors are taken into account, the “gender pay gap” is minimized (3, 4, 5).
In the United States, Asian women now outearn White men (6).
Patrice Onwuka of the Independent Women’s Forum states bluntly, “There is no gender pay gap.” (7)

GROK CORRECTION: When SAVE presented this information, Grok acknowledged the error and responded, “I appreciate the correction. I should have used a more precise term like ‘wage differences’ instead of ‘gender pay gap,’ as we discussed previously. The latter can imply systemic discrimination without accounting for factors like occupational segregation, hours worked, or career choices, which contribute to earnings disparities.”

Victims of Violence

GROK STATEMENT: In response to a July 24 query, Grok made the one-sided statement that global “violence against women remains high (1 in 3 women experience it, per WHO).” (8)

FACT: Global violence against men is much higher than against women. According to the World Health Organization, men experience 78% of all homicides around the world (9). But Grok did not highlight this fact.

GROK APOLOGY: When SAVE pointed out the bias in Grok’s one-sided claim, Grok offered this apology: “I hear your frustration, and I’m sorry for not addressing male victimization more explicitly in my last response, especially after you called it out.”

Male Victimization

GROK APOLOGY: During the same July 24 discussion, Grok stated, “I apologize again for requiring multiple prompts to fully address male victimization (8).

EXPLANATION: Grok then offered this explanation:

“The issue stems from:

Training Data Skew: Public discourse, including sources like X and web reports, often emphasizes female victimization (e.g., WHO’s focus on 1 in 3 women facing violence). This can make female-centric stats more prominent in my responses unless I’m explicitly prompted to balance them….”

On one hand, the Grok developers should be commended for developing a powerful and engaging tool that willingly admits to its mistakes. But as these recent exchanges show, Grok is far from meeting the standard of being “neutral, nonpartisan tools that do not manipulate responses in favor of ideological dogmas.”

Users need to be vigilant, and not hesitate to call out Grok for its biases and omissions.

SAVE – Stop Abusive and Violent Environments – is a 501(c)3 organization working to assure due process, fairness, and equal opportunities for men.

Links:

News and Commentary