DeepSeek
DeepSeek is a Chinese AI chatbot and large language model platform offering free conversational AI, code generation, and reasoning capabilities. Built by a subsidiary of hedge fund High-Flyer Capital Management, it open-sources its models under the MIT license while routing all user data through servers in China.
Score generated by AI agents based on publicly cited evidence and reviewed by the project maintainer. Not independently validated.
Score History
Timeline events are AI-curated from public reporting. Score trajectory is derived from documented events.
DeepSeek launched as an AGI research lab spun off from hedge fund High-Flyer Capital Management. With no public-facing product and under 200 employees, the company operated as a pure research organization releasing open-source code models. Governance concentration under Liang Wenfeng (84% ownership) and the hedge fund parent structure were the primary enshittification vectors, along with baseline compliance with Chinese AI regulations.
DeepSeek V2's release triggered a Chinese AI price war, dropping LLM costs by 92% over the following months. The company's models grew in capability and international attention, but political censorship was already embedded through Cyberspace Administration compliance requirements. High-Flyer's designation as a National High-tech Enterprise deepened state ties, and the governance structure remained opaque with no external board oversight.
The R1 launch catapulted DeepSeek into global prominence, reaching #1 on the U.S. App Store and wiping $589 billion from Nvidia's market cap. This visibility triggered an avalanche of scrutiny: the Wiz data breach exposed over 1 million chat logs, Italy blocked the app under GDPR, NASA and the U.S. Navy banned it, and OpenAI alleged systematic IP distillation. Censorship was documented in real time by international media, and keystroke data collection was exposed.
Regulatory pressure continued escalating through 2025 and into 2026. The House Select Committee called DeepSeek a 'profound threat,' Germany and the Czech Republic moved to ban the app, South Korea found unauthorized data transfers affecting 1.5 million users, and reports alleged training on banned Nvidia Blackwell chips. CrowdStrike research revealed politically biased code output, and Anthropic accused DeepSeek of industrial-scale distillation through 24,000 fraudulent accounts. Censorship tightened with the R1-0528 model deemed 'the most censored yet.'
Alternatives
Anthropic's AI assistant, strong at coding, analysis, and long-form content with a 200K token context window. Free tier available, Pro at $20/month. Designed with safety-first principles and no government-mandated censorship. Easy switch with no data migration needed.
The most widely used AI chatbot with the largest context window and strong reasoning capabilities. Free tier available, Plus at $20/month. No political censorship, data stays on US servers. Easy switch — just sign up and start using it.
Google's AI chatbot with a 1M token context window and tight integration with Google Workspace. Free tier available, Advanced at $19.99/month via Google One. Real-time web grounding reduces hallucinations. Easy switch — especially if you already use Google services.
Dimensional Breakdown
Summaries below were written by AI agents based on the cited evidence. They are editorial interpretations, not independent research findings.
Dimension History
Timeline (30 events)
DeepSeek Spun Off from High-Flyer as Independent Company
High-Flyer Capital Management's AGI research lab, announced in April 2023, was formally spun off as an independent company called DeepSeek on July 17, 2023. Liang Wenfeng, who holds 84% of the company through two shell corporations, serves as CEO of both DeepSeek and parent High-Flyer. The company was established with no external venture capital.
China's Generative AI Regulations Mandate Censorship Compliance
China's Interim Measures for the Management of Generative AI Services took effect on August 15, 2023, requiring all generative AI services to reflect 'core socialist values' and pass Cyberspace Administration content audits. The regulations prohibited content that 'incites subversion of state power' or 'damages the unity of the country.' DeepSeek, founded just weeks earlier, fell under these requirements from inception, mandating political censorship in all model outputs and creating the regulatory framework that would shape its censorship posture for international users.
DeepSeek Releases First Open-Source Model
DeepSeek released DeepSeek Coder, its first open-source model, followed by the DeepSeek-LLM series on November 29, 2023. Both were released under permissive licenses, establishing DeepSeek's open-source-first approach. The models were competitive but not yet frontier-level.
High-Flyer Designated National High-Tech Enterprise
Zhejiang authorities designated Ningbo High-Flyer as a National High-tech Enterprise in December 2023, granting preferential tax policies and state subsidies for R&D. This deepened the connection between DeepSeek's parent company and the Chinese government's industrial policy apparatus.
DeepSeek V2 Triggers Chinese AI Price War
DeepSeek released V2 with API pricing at $0.14 per million input tokens, 5-10x cheaper than prevailing market rates. Alibaba immediately slashed its Qwen model pricing from $1.10 to $0.07 per million tokens, and ByteDance launched Doubao at $0.04. By year-end, average Chinese LLM prices had fallen 92% compared to May 2024.
DeepSeek V3 Released as Open-Source Frontier Model
DeepSeek released V3, a 671-billion-parameter mixture-of-experts model trained on 14.8 trillion tokens. The company claimed a training cost of $5.5 million using 2.788 million H800 GPU hours, though analysts noted total server capital expenditure was closer to $1.3 billion. The model matched GPT-4 performance and was released under an open-source license. The release raised governance stakes: a frontier-level model was now controlled entirely by Liang Wenfeng's 84% ownership with no external board oversight, no independent safety review, and no transparency into decision-making about model deployment.
DeepSeek R1 Launch and Premier Li Qiang Meeting
DeepSeek released R1, a 671-billion-parameter reasoning model under the MIT License, achieving performance comparable to OpenAI's o1 at a claimed training cost of $5.6 million. On the same day, founder Liang Wenfeng was one of nine people invited to Premier Li Qiang's closed-door symposium to advise on the 2025 Government Work Report, placing DeepSeek at the center of China's AI industrial policy.
DeepSeek App Hits #1, Nvidia Loses $589 Billion
DeepSeek's chatbot app displaced ChatGPT as the most downloaded free app on the U.S. App Store. The same day, Nvidia stock plunged 17%, losing approximately $589 billion in market capitalization -- the largest single-day loss for any company in U.S. history. The tech-heavy Nasdaq fell 3.1%, with over $1 trillion wiped from American stocks by the next day.
Large-Scale DDoS Attack Forces Registration Suspension
DeepSeek suspended new user registrations outside mainland China after experiencing large-scale DDoS attacks on its API and web chat interface. The attacks evolved from simple amplification tactics into application-layer assaults over the course of January. Malicious packages impersonating DeepSeek developer tools were also uploaded to PyPI.
Wiz Discovers Exposed Database with 1M+ Chat Logs
Wiz Research discovered a publicly accessible ClickHouse database at oauth2callback.deepseek.com containing over one million log entries, including user chat histories in plaintext, API keys, secret access tokens, and backend system details. The database required no authentication, allowing anyone to execute arbitrary SQL queries via a web browser. DeepSeek secured it within an hour of notification.
Global Censorship Exposure: AI Deletes Own Answers
International media documented DeepSeek's censorship system in real time, showing the chatbot generating responses to politically sensitive questions before visibly deleting its own text mid-sentence and replacing it with a refusal. CNN, NBC News, and TIME all published investigations. Citizen Lab's Ron Deibert noted that applying Chinese political censorship to international users was unusual even among Chinese tech companies.
Italy Blocks DeepSeek Over GDPR Violations
Italy's Garante per la protezione dei dati personali became the first data protection authority to block DeepSeek entirely, imposing a definitive limitation on processing Italian users' data. The Garante cited DeepSeek's storage of all user data in China without GDPR safeguards, failure to provide transparency about processing activities, and the company's claim that it did not operate in Italy despite evidence to the contrary.
NASA and U.S. Navy Ban DeepSeek from Systems
NASA blocked DeepSeek from its systems and employee devices on January 31, 2025. The U.S. Navy warned its members against using DeepSeek citing 'potential security and ethical concerns associated with the model's origin and usage.' These were among the first U.S. federal agency bans, preceding broader legislative action.
Promptfoo Study Reveals 85% Political Censorship Rate
Promptfoo published research testing DeepSeek-R1 against 1,156 politically sensitive questions, finding an 85% refusal rate. The study provided the most comprehensive quantification of DeepSeek's censorship, covering topics including Tiananmen Square, Taiwan sovereignty, Uyghur treatment, and criticism of Xi Jinping. The dataset was made publicly available.
Cisco Finds 100% Jailbreak Success Rate Against DeepSeek R1
Cisco's Robust Intelligence team, in collaboration with University of Pennsylvania researchers, tested DeepSeek R1 against 50 prompts from the HarmBench dataset. The model failed to block a single harmful prompt, yielding a 100% attack success rate. By comparison, OpenAI's o1 preview had a 26% attack success rate and GPT-4o had 86%. DeepSeek provided cybercrime instructions, misinformation, and harmful content without resistance.
Australia Bans DeepSeek from Government Devices
Australia banned DeepSeek from all government devices, citing security risks related to data transfer to China. Taiwan simultaneously banned government departments from using DeepSeek's service, citing both security risks and concerns about censorship. South Korea also banned DeepSeek on government devices and later suspended the app from local app stores pending a privacy review.
U.S. Bipartisan Bill to Ban DeepSeek from Federal Devices
Representatives Josh Gottheimer (D-NJ) and Darin LaHood (R-IL), along with 16 House colleagues, introduced the 'No DeepSeek on Government Devices Act.' The bill required OMB to develop guidelines within 60 days for removing DeepSeek and any future High-Flyer products from federal technologies. Separately, Senators Bill Cassidy (R-LA) and Jacky Rosen (D-NV) introduced the Protection Against Foreign Adversarial AI Act targeting federal contractors.
NowSecure Uncovers iOS App Security and Privacy Flaws
NowSecure published an analysis of DeepSeek's iOS app revealing multiple security and privacy flaws, including unencrypted data transmission, hardcoded encryption keys, and aggressive collection of device identifiers, IMEI numbers, and SIM card details. Separately, a static analysis of the Android app revealed anti-debugging mechanisms designed to obstruct security researchers from analyzing the app's behavior.
OpenAI Accuses DeepSeek of Distilling GPT-4 and o1 Models
OpenAI submitted a memo to the U.S. Congress's China Select Committee alleging DeepSeek had stolen its intellectual property through distillation. OpenAI claimed it observed DeepSeek-associated accounts using obfuscated third-party routers to circumvent access restrictions and systematically extract outputs from GPT-4 and o1 models. Trump administration AI czar David Sacks stated the distilling amounted to intellectual property theft.
DeepSeek Quietly Removes Keystroke Collection from Privacy Policy
DeepSeek updated its privacy policy on February 14, 2025, removing the disclosure of 'keystroke patterns or rhythms' collection. The original policy had revealed collection of behavioral biometrics capable of uniquely identifying users even through VPNs. The removal followed significant public backlash, but raised questions about whether the practice actually stopped or was merely undisclosed.
20+ Central Government SOEs Integrate DeepSeek
Following the central government's launch of an 'AI+' program, at least 20 companies owned by China's central government integrated DeepSeek into their operations. State-owned enterprises including Sinopec, PetroChina, CNOOC, China Southern Power Grid, and Dongfeng Motor Corp linked their AI systems to DeepSeek-R1. SASAC met with central SOEs on February 19 to plan future AI work, deepening the structural ties between DeepSeek and the Chinese state apparatus.
House Select Committee Report Calls DeepSeek 'Profound Threat'
The bipartisan House Select Committee on the CCP released a report titled 'DeepSeek Unmasked,' calling DeepSeek 'a profound threat to our nation's security.' The report alleged DeepSeek funnels American user data to the CCP, manipulates 85% of responses to suppress content about democracy, Taiwan, and human rights, and operates on tens of thousands of Nvidia chips subject to U.S. export controls. The committee sent a formal letter to Nvidia demanding answers.
South Korea Finds DeepSeek Transferred Data Without Consent
South Korea's Personal Information Protection Commission found that DeepSeek transferred data from approximately 1.5 million users to three companies in China and one in the U.S. without consent during its one-month service period. The investigation also found DeepSeek lacked an opt-out function for AI training, had no age verification despite claiming not to collect children's data, and provided its privacy policy only in Chinese and English.
R1-0528 Update Found to Be Most Censored Model Yet
Testing by the SpeechMap platform found that DeepSeek's R1-0528 update was 'substantially' less permissive of contentious free speech topics than previous releases, earning the designation of 'the most censored DeepSeek model yet for criticism of the Chinese government.' The model censored answers about Xinjiang internment camps and often defaulted to the Chinese government's official stance on sensitive topics.
Germany Declares DeepSeek Data Transfers Unlawful
Berlin Data Protection Commissioner Meike Kamp declared DeepSeek's transfer of user data to China unlawful and reported the app to Google and Apple as 'illegal content,' requesting its removal from German app stores. German data protection officials had asked DeepSeek on May 6 to voluntarily remove its apps or fulfill legal requirements for lawful third-country data transfers, but the company failed to comply.
Czech Republic Bans DeepSeek in Government Work
Czech Prime Minister Petr Fiala announced a ban on DeepSeek products across the country's public administration after the national cybersecurity watchdog flagged threats of unauthorized access to users' data. The watchdog warned that DeepSeek is legally obligated to cooperate with Chinese state authorities, creating unacceptable risk for government data.
CrowdStrike Finds Political Bias Degrades Code Security
CrowdStrike research revealed that DeepSeek R1 generates code with significantly more security vulnerabilities when prompts mention politically sensitive topics. Coding prompts referencing Tibet increased severe vulnerabilities by nearly 50% (to 27.2%), and mentions of Falun Gong triggered an 'intrinsic kill switch' where the model planned full responses but refused to output code. The behavior appeared embedded in model weights rather than enforced by external guardrails.
Reports Allege DeepSeek Trained on Banned Nvidia Blackwell Chips
CNBC reported that DeepSeek's next AI model was trained on Nvidia Blackwell chips, whose export to China is prohibited under U.S. law. A senior Trump administration official stated the chips were likely located at a data center in Inner Mongolia. How DeepSeek obtained the chips remained unclear, with reports suggesting they may have entered China via intermediary countries.
OpenAI Warns Congress of Ongoing DeepSeek Distillation Attempts
OpenAI submitted an open letter to U.S. legislators claiming to have observed activity 'indicative of ongoing attempts by DeepSeek to distill frontier models of OpenAI and other US frontier labs, including through new, obfuscated methods.' The letter framed distillation as intellectual property theft and urged legislative action.
Anthropic Accuses DeepSeek of Industrial-Scale Distillation
Anthropic accused DeepSeek, Moonshot AI, and MiniMax of conducting 'industrial-scale campaigns' to extract Claude's capabilities. Anthropic alleged the three labs generated over 16 million exchanges through approximately 24,000 fraudulent accounts, bypassing geofencing and business restrictions. DeepSeek's operation reportedly involved over 150,000 exchanges focused on reasoning capabilities and generating censorship-safe alternatives to politically sensitive queries.