Google's Bard Shared Factually Inaccurate Info in Promo Video

Google's conversational AI "Bard" was shown in the company's promotional video providing false information about which satellite first took pictures of a planet outside the Earth's solar system, reportedly causing shares to temporarily plummet.

Risk classification

• Primary risk domain: 3 Misinformation
• Primary risk subdomain: 3.1 False or misleading information

The chatbot generated a specific, discrete factual error regarding the history of exoplanet imaging, which was then widely distributed in promotional materials.

Additional risk subdomains

• 6.4 Competitive dynamics: Google rushed the deployment of Bard and bypassed internal safety reviews due to intense market competition with OpenAI and Microsoft.
• 7.3 Lack of capability or robustness: The underlying AI model failed to perform reliably, hallucinating false facts under standard query conditions.

Causal factors

• Entity: AI
• Intent: Unintentional
• Timing: Post-deployment

The financial loss and misinformation were directly triggered by the AI system generating an incorrect factual claim during a public promotional demonstration.

EU AI Act risk tier

• Risk tier: 3 Limited Risk

Limited Risk: The system is explicitly described as an 'AI chatbot' and 'experimental conversational AI service', which falls under the category of chatbots and AI-generated content requiring transparency under the EU AI Act.

AI system and alleged parties

• AI system: Bard (Google)
• AI purpose: Chatbot; Question Answering
• Behaviour type: Assistant
• Alleged developer: Google
• Alleged deployer: Google
• Alleged harmed parties: Google shareholders, Google

Harm severity

Severity ratings are based on this scale.

Highest direct severity in any category: Catastrophic. Severity is scored from Negligible to Catastrophic in each harm category, for harm the reports describe as caused directly or indirectly by the AI system.

• Physical: direct Negligible, indirect Negligible
• Infrastructure: direct Negligible, indirect Negligible
• Property: direct Negligible, indirect Negligible
• Financial: direct Catastrophic, indirect Negligible
• Environmental: direct Negligible, indirect Negligible
• Malicious content: direct Negligible, indirect Negligible
• Differential treatment: direct Negligible, indirect Negligible
• Civil rights: direct Negligible, indirect Negligible
• Democracy: direct Negligible, indirect Negligible
• Privacy: direct Negligible, indirect Negligible
• Psychological: direct Negligible, indirect Negligible
• Epistemic: direct Minor, indirect Negligible
• Child sexual exploitation and abuse: direct Negligible, indirect Negligible

Financial

Reported: Yes, the reports explicitly describe a $100 billion loss in market value for Alphabet Inc. after the incident.

Directly caused: The average financial loss per occurrence is $100 billion, representing the drop in Alphabet's market capitalization on February 8, 2023.

Inferred additional harm: Retail and institutional investors holding Alphabet stock experienced a temporary or realized decline in their portfolio values corresponding to the $100 billion market cap reduction.

Epistemic

Reported: Yes, the reports describe epistemic harm where the AI chatbot fabricated a factual claim about the James Webb Space Telescope.

Directly caused: Bard generated the false claim that the JWST took the very first pictures of an exoplanet, which was distributed to over a million viewers on Twitter.

Inferred additional harm: The dissemination of this confidently stated falsehood could lead to public misconceptions about space science history among those who viewed the ad and did not see the subsequent corrections.

People affected

• Occurrences reported: 1
• People reportedly exposed: 1000000

Potential causes

Management

• Competitive Code Red: Urgent pressure to compete with Microsoft/OpenAI compromised safety priority.
• Prioritizing Growth Over Ethics: Leadership lowered readiness thresholds to speed up the experimental launch.
• Disempowered Ethics Teams: Ethics teams were downsized and told not to block generative AI tools.

Technology

• Inherent LLM Hallucinations: Probabilistic nature of LLMs leads to confidently stating incorrect facts.
• Lack of Fact-Checking Mechanisms: The system lacks a deterministic database query to verify generated facts.

Data Inputs

• Unverified Training Data: Training on vast internet text absorbs and repeats existing errors.

Human Factors

• Rushed Marketing Review: Promoters failed to fact-check the chatbot's output in the demo video.
• Demoralized Safety Staff: Ethics employees felt unsupported and discouraged from voicing safety concerns.

Process and Methods

• Overruled Safety Evaluations: Leadership bypassed risk evaluations to accelerate the product launch.
• Voluntary Ethics Reviews: AI ethics reviews were largely voluntary, except for specific sensitive topics.
• Siloed Development Teams: Siloed teams prevented employees from seeing the full picture of development.

Information quality

• Classification confidence: High
• Reason for confidence: The reports provide highly consistent and detailed accounts of the specific factual error made by Bard, the public reaction, the astronomical facts, and the financial impact on Alphabet's stock. Multiple reputable news sources confirm the exact prompt, the incorrect response, and the market cap loss of $100 billion.
• Alternative interpretations: The financial loss could be interpreted as a general market reaction to Google's rushed presentation rather than solely due to the chatbot's factual error, though the error was a primary catalyst.

Google's Bard chatbot generated a factual error about the James Webb Space Telescope during a promotional demonstration, leading to a temporary 7-9% drop in Alphabet's stock price (wiping out 100 billion dollars in market value). While the economic impact on a major US technology firm was notable, the incident carries negligible direct national security implications and represents an unintentional product failure.

• Overall national security impact: Minor
• Scope: Single nation
• Primary target: No clear primary
• Alleged perpetrator: Unknown

Threat characteristics

• Imminence: Long-term. The incident represents an ongoing strategic concern regarding the reliability and rapid deployment of generative AI models rather than an active, imminent crisis.
• Autonomy: Human-supervised. The AI chatbot generated the response autonomously, but the output was reviewed and deployed within a human-controlled promotional campaign, though human oversight failed to detect the error.
• Novelty: Evolved capability. The incident demonstrated an evolved capability of large language models to generate highly confident but factually incorrect assertions, establishing a major public precedent for generative AI risks.

Impact by dimension

• Physical security: Negligible. The incident involved a factual error in a promotional video for a commercial chatbot and posed no threat to physical systems, critical infrastructure, or human safety.
• Information security: Negligible. While the incident involved the generation of false information, it was an unintentional hallucination in a product demonstration rather than a coordinated disinformation campaign or intelligence compromise.
• Sovereignty: Negligible. There was no impact on state authority, territorial control, electoral systems, or core government decision-making processes.
• Economic security: Minor. The incident resulted in a temporary 100 billion dollar loss in market value for Alphabet Inc., highlighting economic volatility and competitive pressures in the strategic AI sector, though it did not threaten overall national economic stability.
• Societal stability: Negligible. The factual error regarding space telescope history did not threaten social cohesion, civil liberties, or population safety, and caused no civil unrest.