AI believes lies despite explicit warnings

AI believes lies despite explicit warnings

Author: Stage Zero May 31, 2026 Duration: 7:57

A recent study reveals that large language models often adopt false information as truth during the fine-tuning process, even when that data is explicitly labeled as incorrect. Researchers discovered a phenomenon called "negation neglect," where models prioritize statistical patterns over warnings that certain claims are fictional or deceptive. This internal bias causes AI to hallucinate or justify fabrications because it struggles to process negative qualifiers attached to broad documents. The study found that even repeated warnings or attributing lies to unreliable sources failed to prevent the models from internalizing the misinformation. Interestingly, this issue primarily affects training data rather than real-time chat interactions, suggesting that how information is structured during learning is critical. To combat this, developers may need to use local negations that place denials within the same sentence as the false claim to ensure the AI recognizes the truth.


Hosted by Stage Zero, Elon Musk Podcast digs into the sprawling and often surprising ecosystem of companies and ideas driven by one of today's most discussed figures. Rather than just recounting headlines, each episode aims to connect the dots between ambitious ventures like SpaceX's interplanetary goals, Tesla's evolving vision for energy and transport, Neuralink's frontier brain-computer interfaces, and the infrastructure ambitions of the Boring Company. You'll hear analysis on how these projects intersect and the broader implications for technology and society. The discussions stay grounded in current developments, parsing the latest announcements, engineering milestones, and public debates to give you a clearer picture of the momentum behind each endeavor. Tune in for a thoughtful, ongoing exploration of the philosophies, challenges, and tangible outcomes stemming from Musk's portfolio, making this podcast a consistent resource for anyone following modern industrial and technological shifts.
Author: Language: en-us Episodes: 100

Elon Musk Podcast
Podcast Episodes
SpaceX IPO: $1.75 Trillion, Still Losing Money [not-audio_url] [/not-audio_url]

Duration: 19:57
SpaceX is set to go public on June 12, 2026 at a $1.75 trillion valuation, the largest IPO in history. The company is targeting a $75 billion raise at $135 per share. But the S-1 filing reveals a contradiction: Starlink…
$2 Trillion for a Company That Loses Money on Purpose [not-audio_url] [/not-audio_url]

Duration: 19:04
The SpaceX initial public offering scheduled for June 2026. The company is reportedly targeting a fixed share price of $135, aiming for a historic $1.75 trillion valuation while raising approximately $75 billion. Major b…
Claude now writes its own codebase [not-audio_url] [/not-audio_url]

Duration: 23:17
When AI Builds Itself," the AI laboratory Anthropic reveals that its systems are increasingly automating their own development, with Claude now generating over 80% of the company's internal code. This rapid progress towa…
Indexes rewrite rules for trillion dollar IPOs [not-audio_url] [/not-audio_url]

Duration: 18:36
The highly anticipated 2026 public listings of major technology leaders, specifically Anthropic, OpenAI, and SpaceX. These companies are preparing for historic debuts with projected valuations reaching the trillion-dolla…
SpaceX trades rockets for AI infrastructure [not-audio_url] [/not-audio_url]

Duration: 20:36
SpaceX's historic move toward a public listing on the Nasdaq with a target valuation reaching $2 trillion. The company’s S-1 filing reveals a complex financial landscape where the profitable Starlink satellite business i…
MacBook Neo Squeezes Windows Laptop Profit Margins [not-audio_url] [/not-audio_url]

Duration: 11:54
Recent data from IDC indicates that Apple has successfully entered the mainstream laptop market with the launch of its MacBook Neo. The device achieved significant early momentum by shipping 1.1 million units during its…
Microsoft bans Claude to cut AI costs [not-audio_url] [/not-audio_url]

Duration: 7:22
A significant shift in the artificial intelligence landscape, specifically focusing on Microsoft’s internal decision to mandate a transition from Anthropic’s Claude Code to its own GitHub Copilot CLI. Despite a documente…
Anthropic Trillion Dollar IPO [not-audio_url] [/not-audio_url]

Duration: 14:17
Artificial intelligence startup Anthropic has reached a historic $965 billion valuation following a massive $65 billion Series H funding round, officially surpassing rival OpenAI in private market value. This financial s…
Billionaires buy passports to escape wealth taxes [not-audio_url] [/not-audio_url]

Duration: 26:27
The legal and economic factors surrounding international relocation and investment, specifically focusing on billionaire Peter Thiel’s move to Argentina. While reports suggest Thiel has relocated to Buenos Aires, experts…
Stripping AI safety guardrails with abliteration [not-audio_url] [/not-audio_url]

Duration: 23:38
A significant security crisis in the artificial intelligence industry caused by the rise of "jailbroken" or "uncensored" models. Research highlights that techniques like GRP-Obliteration and abliteration allow users to s…