AI-generated images raise challenges in combating misinformation

The Los Angeles fires were raging when Mark Zuckerberg, CEO of Meta, announced the removal of third-party fact-checkers from the social media platform. Around the same time, an image of a house untouched by the surrounding inferno began circulating online, capturing public attention. 

However, it didn’t take long for experts to confirm that the image was AI-generated — a clear example of how Generative AI (GenAI) is reshaping the landscape of misinformation.

GenAI, which allows users to create realistic images and content with minimal technical knowledge, has opened the door to unprecedented creativity. At the same time, it has complicated efforts to verify authenticity. In fact, the World Economic Forum’s “Future of Jobs Report 2025” lists graphic design as one of the professions most affected by artificial intelligence, as automation threatens to disrupt traditional creative roles.

While AI tools have made content creation more accessible, they have also fueled the spread of fake visuals, forcing individuals and organizations to grapple with new challenges in identifying and combating disinformation. With Meta’s decision to eliminate third-party fact-checkers, the burden of verifying content now falls increasingly on users.

The dangers of AI-generated images

“Without proper disclosure, AI-generated images could be extremely dangerous, as they may be used to spread fake news and black propaganda,” Gerando Bonganay, AI architect at IBM Philippines, told Back End News in an email interview. “They can perpetuate manipulative biases that incite social unrest.”

The risks associated with AI-generated content are multifaceted. Sophisticated algorithms can produce visuals indistinguishable from reality, enabling bad actors to create convincing yet false narratives. These images can influence public opinion, undermine trust, and sow confusion.

AI-generated images are particularly problematic during crises, such as natural disasters or elections, when accurate information is critical. The speed at which these visuals spread on social media platforms further complicates the issue, often outpacing efforts to debunk them.

The evolving sophistication of AI

Artificial intelligence has been around for years, but recent advancements have significantly enhanced its capabilities. AI-generated content is now so sophisticated that traditional methods of verification struggle to keep pace.

Before AI became mainstream, creators relied on tools like watermarks to protect copyrighted material. However, AI-powered tools can now erase watermarks, making it easier to manipulate and repurpose content. Device manufacturers even advertise AI-enhanced editing features as selling points, further normalizing the technology.

“There is no fool-proof way to spot AI-generated images,” said John Shier, field CTO and senior manager at Sophos. He explained that while some initiatives aim to improve authentication, they face significant hurdles.

According to Shier, efforts like The Coalition for Content Provenance and Authenticity (C2PA) are working to develop standards for verifying digital content. These include embedding digital watermarks and hidden patterns in AI-generated images to establish their origins.

“Researchers are also exploring machine learning-based detection techniques to recognize anomalies in color, lighting, shadows, and textures,” Bonganay said.

However, even advanced tools have limitations. Watermarks can be removed, and metadata — the digital information embedded in images — can be altered, reducing the reliability of these methods.

Metadata and reverse image searches

Metadata, such as EXIF data, can reveal valuable information about an image’s origins, including the camera model, settings, and timestamps. This data is often missing or intentionally removed from AI-generated content.

“AI-generated images often lack intact metadata,” Bonganay explained. “If an image has been edited to remove or alter metadata, our ability to identify it as AI-generated becomes compromised.”

Reverse image searches, such as Google’s image search feature, can also help verify authenticity. By comparing an image to similar visuals online, users can determine whether it has been manipulated or fabricated.

However, Shier cautioned against relying solely on reverse image searches. “They shouldn’t be the only method used,” he said, emphasizing the importance of a multi-pronged approach to verification.

Spotting AI-generated images

For everyday users, identifying AI-generated content requires a combination of observation and skepticism. Bonganay and Shier offered practical tips for spotting potential red flags in images:

• Unnatural facial features: Look for symmetrical faces, overly idealized structures, or proportions that seem too perfect.
• Inconsistent backgrounds: Examine the lighting, textures, and composition of the background for signs of manipulation.
• Composition errors: Pay attention to objects that don’t match their surroundings or appear unnaturally posed.
• Digital artifacts: Watch for pixelation, over-smoothing, or inconsistent resolution, which may indicate AI manipulation.

Shier also recommended using the SIFT method: Stop, Investigate the claim, Find better coverage, and Trace sources to their original context. This systematic approach encourages users to pause and verify content before sharing it.

The role of critical thinking

In the absence of third-party fact-checkers, critical thinking becomes a crucial tool for combating disinformation. Bonganay stressed the importance of skepticism when encountering sensational or provocative content online.

“When you see something that seems too good or bad to be true, fact-check it with reputable sources,” he said. “By being critical thinkers and seeking out multiple perspectives, we can stay informed and distinguish facts from falsehoods.”

The challenges posed by AI-generated content highlight the need for ongoing innovation in detection and verification technologies. While tools like digital watermarks, metadata analysis, and reverse image searches offer some solutions, they are not perfect.

For now, individuals must take responsibility for verifying the information they consume and share. By combining technology with critical thinking, society can better navigate the complexities of the digital age and uphold the truth in an era shaped by AI.