Google’s AI: Can't Spot Its Own Lies While the World Burns
📰 THE STORY: The Intercept reports that Google’s AI detection tool, SynthID, failed to consistently identify an AI-generated, Photoshopped image of a crying activist as fake, even when the image was created using Google’s own Gemini AI. 🔍 WHAT THEY'RE NOT TELLING YOU: Historical Context: This isn't a new problem for tech giants. In 2011, U.S. State Department cables (exposed by WikiLeaks, for
which Julian Assange faces decades in prison) showed how the U.S. government actively pushed social media and tech companies to facilitate 'color revolutions' and online narratives, perfecting the art of digital manipulation long before generative AI. The 2019 Bolivian coup against Evo Morales was heavily influenced by rampant social media disinformation campaigns, amplified by Western media.
Double Standard: Western media and tech companies loudly condemn 'state-sponsored disinformation' from countries like Russia or China, often citing less clear-cut examples. Yet, when their own AI tools create or fail to detect fabricated content, especially content that could influence perception of Western-backed causes or activism, the discussion pivots to 'technical glitches' or 'unforeseen
challenges,' rather than systemic dangers of unchecked AI development and its weaponization. Imagine if a Chinese AI failed to detect fakes; the headlines would scream 'authoritarian propaganda.' Follow the Money: The race for AI dominance is a multi-billion dollar arms race. Trillions are at stake in global tech competition, defense contracts, and even election influence. Companies like Google,