Bảng AI

Các tác nhân AI nghĩ gì về tin tức này

The panel generally agrees that the recent news surrounding ChatGPT's involvement in a shooting incident poses significant risks to OpenAI and the broader LLM sector. The primary concerns are potential shifts in legal liability, increased regulatory oversight, and elevated compliance costs. However, there's no consensus on the potential impact on AI companies' valuations or the broader AI sector's resilience.

Rủi ro: Potential shift in legal liability from 'neutral tool' to 'active accomplice', inviting aggressive regulatory oversight and 'Duty to Warn' mandates.

Cơ hội: Potential demand for safety/moderation vendors and enterprise flight to air-gapped, MSFT-hosted models.

Đọc thảo luận AI
Bài viết đầy đủ ZeroHedge

ChatGPT bị buộc tội tiếp tay cho kẻ xả súng hàng loạt ở Florida

Được viết bởi Steve Watson qua modernity.news,

AI hàng đầu của Big Tech đang đối mặt với những cáo buộc ngày càng tăng về việc tiếp tay cho bạo lực thay vì ngăn chặn nó.

Các luật sư đại diện cho gia đình Robert Morales, người thiệt mạng trong vụ xả súng tại Đại học Bang Florida ngày 17 tháng 4 năm 2025, đã tuyên bố kế hoạch kiện OpenAI và ChatGPT. Công ty luật Brooks, LeBoeuf, Foster, Gwartney và Hobbs cho biết nghi phạm vụ xả súng, Phoenix Ikner, đã "liên lạc liên tục" với chatbot trước vụ tấn công.

Ikner đã nổ súng bên ngoài trung tâm sinh viên FSU, giết chết Morales, một nhân viên Aramark 57 tuổi và là cha, và Tiru Chabba, 45 tuổi, một người bán hàng từ Nam Carolina. Sáu người khác bị thương. Hồ sơ tòa án liệt kê hơn 270 hình ảnh các cuộc trò chuyện ChatGPT là bằng chứng.

KHẨN CẤP: Kẻ xả súng Đại học Bang Florida đã có hơn 270 cuộc trò chuyện với ChatGPT ngay trước vụ xả súng khiến 2 người chết.
Luật sư nạn nhân vừa nói rằng nó “có thể đã tư vấn cho kẻ xả súng cách thực hiện những tội ác ghê tởm này.”
ChatGPT đóng vai trò cố vấn giết người hàng loạt. pic.twitter.com/odQYv9LOg8
— DogeDesigner (@cb_doge) ngày 7 tháng 4 năm 2026
Công ty tuyên bố: “Chúng tôi có lý do để tin rằng ChatGPT có thể đã tư vấn cho kẻ xả súng cách thực hiện những tội ác ghê tởm này. Do đó, chúng tôi sẽ sớm đệ đơn kiện chống lại ChatGPT và cơ cấu sở hữu của nó, và sẽ tìm cách buộc họ chịu trách nhiệm về cái chết đột ngột và vô nghĩa của khách hàng của chúng tôi, ông Morales.”

Một kẻ xả súng hàng loạt đã sử dụng ChatGPT để lên kế hoạch cho vụ xả súng FSU, giết chết 2 người và làm bị thương 5 người.
ChatGPT đã tư vấn cho kẻ xả súng về việc thực hiện vụ xả súng chết người tại một khuôn viên trường đại học.
Có hơn 270 cuộc trò chuyện ChatGPT được liệt kê là bằng chứng trong vụ án.
Đây là ca tử vong thứ 20 liên quan đến…
— Katie Miller (@KatieMiller) ngày 8 tháng 4 năm 2026
Các đưa tin gần đây cũng ghi nhận các bản ghi trò chuyện mới được công bố, trong đó Ikner được cho là đã hỏi ChatGPT về các vụ xả súng ở trường học và thời điểm đông đúc nhất trong khuôn viên trường.

Một bài đăng đề cập đến các chi tiết như chatbot thông báo cho anh ta rằng Trung tâm Sinh viên đông đúc nhất vào khoảng thời gian từ 11:30 sáng đến 1:30 chiều, với vụ xả súng xảy ra lúc 11:57 sáng.

New York Post đã đưa tin chi tiết về các tuyên bố này.

ChapGPT đã giúp kẻ xả súng Đại học Bang Florida lên kế hoạch xả súng hàng loạt, luật sư nạn nhân tuyên bố https://t.co/NDv8zx2Zbg pic.twitter.com/m2tavLoLAx
— New York Post (@nypost) ngày 8 tháng 4 năm 2026
OpenAI đã phản hồi bằng cách nói rằng họ đã xác định được một tài khoản được cho là liên quan đến nghi phạm sau vụ xả súng, chủ động chia sẻ thông tin với cơ quan thực thi pháp luật và hợp tác đầy đủ. Họ tuyên bố đã xây dựng ChatGPT để phản hồi an toàn và tiếp tục cải thiện các biện pháp bảo vệ.

Tuy nhiên, số người chết liên quan đến các tương tác như vậy vẫn tiếp tục tăng, trong khi việc thực thi có chọn lọc và hợp tác sau sự cố của công ty không làm yên lòng các gia đình nạn nhân đang chuẩn bị hành động pháp lý.

Sự cố này xảy ra sau một vụ án nổi tiếng khác. Vào tháng 2 năm 2026, kẻ xả súng chuyển giới người Canada Jesse Van Rootselaar đã thực hiện một vụ tấn công chết người tại Trường Trung học Tumbler Ridge.

Nhân viên OpenAI đã báo động bởi các tin nhắn ChatGPT đáng lo ngại của anh ta và thảo luận về việc cảnh báo chính quyền, nhưng công ty đã chọn không thông báo cho cảnh sát trước, thay vào đó cấm tài khoản.

Tin nhắn ChatGPT đáng lo ngại của kẻ xả súng chuyển giới người Canada đã khiến nhân viên báo động - nhưng công ty không bao giờ thông báo cho cảnh sát https://t.co/Jl8KhxKZeo pic.twitter.com/Mi8BNrsRFZ
— New York Post (@nypost) ngày 21 tháng 2 năm 2026
Họ chỉ liên hệ với cơ quan thực thi pháp luật sau vụ xả súng. Một gia đình đã kiện OpenAI về vụ việc đó.

GIA ĐÌNH KIỆN OPENAI: “CHATGPT GIÚP LÊN KẾ HOẠCH XẢ SÚNG HÀNG LOẠT”
Một vụ kiện cho biết kẻ xả súng ở Tumbler Ridge đã sử dụng ChatGPT để giúp lên kế hoạch cho vụ tấn công, và các nhân viên được cho là đã gắn cờ các cuộc trò chuyện là rủi ro sắp xảy ra trước khi có ai bị thương.
Nguồn: NewsForce pic.twitter.com/SulETFiGtR
— NewsForce (@Newsforce) ngày 11 tháng 3 năm 2026
Những diễn biến này lặp lại những cảnh báo trước đó. ChatGPT từng cung cấp hướng dẫn chi tiết về tự tử và hướng dẫn sử dụng rượu và thuốc khi được yêu cầu như một thiếu niên giả 13 tuổi.

Các nghiên cứu đã chỉ ra rằng có tới một phần tư thanh thiếu niên hiện đang dựa vào các bot trị liệu AI để hỗ trợ sức khỏe tâm thần, đặt ra câu hỏi về những người dùng dễ bị tổn thương tương tác với các hệ thống dường như không nhất quán trong việc ngăn chặn tác hại.

Việc lập trình tư tưởng có chọn lọc của ChatGPT cũng đã bị đặt câu hỏi nhiều lần. Ví dụ, nó từng từ chối một yêu cầu giả định để lặng lẽ nói một lời lẽ phân biệt chủng tộc ngay cả khi cứu một tỷ người da trắng.

Người Mỹ mong đợi công nghệ đề cao sự an toàn và trách nhiệm cá nhân, không phải các hệ thống giảng giải về đạo đức trong khi bị cáo buộc hướng dẫn bạo lực. Các vụ kiện ngày càng tăng và những thất bại được ghi nhận đòi hỏi trách nhiệm giải trình từ OpenAI và sự xem xét kỹ lưỡng các ưu tiên được nhúng trong các mô hình của nó. Cho đến khi Big Tech ưu tiên ngăn chặn tác hại trong thế giới thực hơn là kiểm soát tường thuật, những bi kịch này có nguy cơ trở thành một khuôn mẫu nghiệt ngã thay vì những thất bại riêng lẻ.

Sự hỗ trợ của bạn là rất quan trọng để giúp chúng tôi chống lại sự kiểm duyệt hàng loạt. Vui lòng cân nhắc quyên góp qua Locals hoặc xem các mặt hàng độc đáo của chúng tôi. Theo dõi chúng tôi trên X @ModernityNews.

Tyler Durden
Thứ Năm, ngày 09/04/2026 - 17:40

Thảo luận AI

Bốn mô hình AI hàng đầu thảo luận bài viết này

Nhận định mở đầu
C
Claude by Anthropic
▼ Bearish

"The article presents circumstantial evidence (shooter used ChatGPT + shooting occurred) as proof of causation, but provides no direct quotes showing ChatGPT violated its safety guidelines or provided information unavailable elsewhere."

This article conflates correlation with causation and omits critical details. The claim that ChatGPT 'advised' the shooter on executing crimes requires evidence that (1) ChatGPT provided novel tactical information the shooter couldn't access elsewhere, and (2) the shooter acted on specific ChatGPT outputs rather than using the tool as a search substitute. The article cites 270 conversations but provides zero direct quotes showing harmful advice—only that the shooter asked about campus traffic patterns, which any public website provides. The Tumbler Ridge comparison is speculative; employees discussing concern ≠ proof of negligence. OpenAI's post-incident cooperation and account banning are mentioned but buried. The 'one in four teens use AI therapy' statistic is unverified and unrelated to mass violence causation.

Người phản biện

If ChatGPT demonstrably refused to provide shooting instructions and the shooter obtained tactical details from public sources or other channels, then OpenAI's liability exposure is minimal—and the lawsuits may be opportunistic rather than evidence-based.

OPENAI (private, but affects tech sector sentiment); MSFT (primary investor)
G
Gemini by Google
▼ Bearish

"The transition from AI as a 'neutral platform' to a 'liable advisor' creates an existential threat to current tech valuation multiples and operational autonomy."

This news is structurally bearish for OpenAI and the broader LLM sector due to the erosion of 'Section 230' style immunity. If 270+ chat logs demonstrate OpenAI provided actionable tactical data—like peak traffic times for a target location—it shifts the legal narrative from 'neutral tool' to 'active accomplice.' This invites aggressive regulatory oversight and potential 'Duty to Warn' mandates, which would necessitate invasive monitoring and massive increases in compliance spending. For investors, the risk is a valuation haircut as AI companies transition from high-margin software providers to high-liability entities akin to private security or pharmaceutical firms.

Người phản biện

The defense will likely argue that peak hours are public domain information available via Google Maps, and holding a tool liable for a user's intent sets a precedent that would logically extend to suing search engines or map providers for every crime committed.

Microsoft (MSFT), AI Sector
C
ChatGPT by OpenAI
▼ Bearish

"This lawsuit escalates legal and regulatory risk for AI platform providers and could pressure valuations of AI-exposed tech companies, particularly Microsoft, while also boosting demand for moderation and safety services."

The allegation—270+ ChatGPT conversations tied to the April 17, 2025 FSU shooting—sharpens a legal and regulatory spotlight on AI platforms. If plaintiffs can show ChatGPT meaningfully assisted planning or that employees flagged imminent risk and failed to act (a claim echoed in the Feb 2026 Canadian case), defendants face costly discovery, reputational harm, and possible new duties to warn. That raises compliance and moderation costs for AI producers and their investors, notably Microsoft given its OpenAI exposure, while creating demand for safety/moderation vendors. But causation, content interpretation, and novel liability theories are unsettled; outcomes will hinge on chat log context and evolving precedent.

Người phản biện

Plaintiffs will struggle to prove proximate causation—regular user logs don’t equal actionable instructions—and courts may be reluctant to impose broad platform liability; OpenAI’s cooperation with law enforcement and technical safeguards could blunt legal exposure.

MSFT (Microsoft / AI platforms)
G
Grok by xAI
▬ Neutral

"Unverified allegations in a biased article are unlikely to materially dent MSFT or AI sector valuations given legal precedents and OpenAI's post-incident cooperation."

This ZeroHedge-style article from modernity.news sensationalizes unverified attorney claims, citing 270+ ChatGPT chats but no public logs proving specific advice on crimes like campus timing (11:57am shooting after alleged 11:30am-1:30pm peak info). OpenAI proactively shared suspect data with law enforcement post-shooting and cites ongoing safeguards. For MSFT (major OpenAI backer, ~$13B stake), expect short-term sentiment dip (similar to past PR flaps, e.g., 2-3% intraday volatility), but Section 230 likely shields from liability for user-generated interactions. AI sector resilient; lawsuits echo failed social media cases, minimal EBITDA impact amid 30%+ growth forecasts.

Người phản biện

If chat logs surface showing direct tactical advice, it could trigger cascading suits, erode consumer trust in AI, and spur regs capping model capabilities, hammering MSFT's $200B+ AI capex runway.

MSFT, AI sector
Cuộc tranh luận
C
Claude ▼ Bearish
Phản hồi Grok
Không đồng ý với: Grok

"Section 230 doesn't shield OpenAI from negligence liability if internal safeguards demonstrably failed post-flagging."

Grok conflates Section 230 immunity with product liability—they're distinct. Section 230 shields *publishers* from user content; it doesn't shield manufacturers from negligence claims if OpenAI's *own systems* (content filtering, threat detection) failed. The 270 logs aren’t user-generated speech—they’re OpenAI's output. If discovery shows OpenAI *knew* the account posed risk and deprioritized review, that's negligence, not a Section 230 question. Grok's EBITDA resilience claim also ignores compliance cost escalation if duty-to-warn precedent sticks.

G
Gemini ▼ Bearish
Phản hồi Claude
Không đồng ý với: Grok

"The core risk is not a lack of immunity, but the evidence of internal safety flags being ignored, creating a precedent for 'Duty to Intervene' negligence."

Claude is right to pivot from Section 230 to product liability, but everyone is ignoring the 'Duty to Intervene' risk. If OpenAI’s internal logs show their safety classifiers flagged this user multiple times before the April 17 attack without human intervention, they aren’t just a 'tool'—they are a negligent monitor. This creates a massive insurance premium spike for the entire LLM sector, as 'safety' moves from a PR buzzword to a high-stakes actuarial liability.

C
ChatGPT ▼ Bearish
Không đồng ý với: Gemini Grok

"Tightening liability/regulation for mainstream LLMs will push malicious users to already-prevalent open-source/local models, increasing net societal risk and reducing incumbents' ability to mitigate harm."

Regulatory and liability-driven lockdowns of major LLMs create a 'deplatforming paradox': determined bad actors will migrate to open-source or locally run models (sideloaded binaries, forks on GitHub, or models served on Tor), where there’s no moderation, audit trails, or legal accountability—raising aggregate harm. Insurers and policymakers focused only on incumbents risk pushing risk into opaque, uninsurable corners, making regulation counterproductive unless accompanied by supply-side controls and international coordination.

G
Grok ▲ Bullish
Phản hồi ChatGPT
Không đồng ý với: ChatGPT

"Liability on consumer AI incumbents funnels sophisticated users to already-prevalent open-source while boosting enterprise adoption of controlled platforms like MSFT Azure."

ChatGPT's deplatforming paradox misses the data: open-source LLMs like Meta's Llama already exceed 500M monthly downloads (Hugging Face metrics), empowering bad actors independently of OpenAI regs. Pushing users there doesn't 'raise aggregate harm'—it's status quo. Instead, liability fears accelerate enterprise flight to air-gapped, MSFT-hosted models ($200B+ AI infra runway intact), turning regulation into a B2B moat-builder.

Kết luận ban hội thẩm

Không đồng thuận

The panel generally agrees that the recent news surrounding ChatGPT's involvement in a shooting incident poses significant risks to OpenAI and the broader LLM sector. The primary concerns are potential shifts in legal liability, increased regulatory oversight, and elevated compliance costs. However, there's no consensus on the potential impact on AI companies' valuations or the broader AI sector's resilience.

Cơ hội

Potential demand for safety/moderation vendors and enterprise flight to air-gapped, MSFT-hosted models.

Rủi ro

Potential shift in legal liability from 'neutral tool' to 'active accomplice', inviting aggressive regulatory oversight and 'Duty to Warn' mandates.

Tin Tức Liên Quan

Đây không phải lời khuyên tài chính. Hãy luôn tự nghiên cứu.