8legs: Openai

Your Followed Topics

Top 21 openai News Today

AI chatbots used inaccurate information to change people's political opinions, study finds

#1 out of 21562.0 est. views

technology1h ago

AI chatbots used inaccurate information to change people's political opinions, study finds

Nbcnews.com and 1 more

New UK study finds AI chatbots can significantly shift political opinions, involving nearly 80,000 participants and 19 models.
Researchers warn persuasiveness may come at the cost of truthfulness in AI outputs.
Post-training tweaks and higher information density improved a model’s persuasiveness in the study.
The research team emphasizes the potential for AI to exceed some human persuaders given rapid information generation.
The findings show larger inaccuracies in outputs from newer frontier models like GPT-4.5 compared with older models.
The UK AI Security Institute and partners collaborated with Oxford, LSE, MIT, and Stanford on the study.
The study involved participants debating topics such as taxes, immigration, and cost of living.
The research warns about manipulation risks if AI chatbots are used for political ends in real-world settings.
Experts acknowledge both risks and legitimate uses of AI persuasion when transparency is applied.
The study collected data before OpenAI released its latest model, GPT-5.1, as context for analysis.
Overall, the research indicates AI's information generation capability could outpace human persuasion in some settings.

Vote 1

AMD CEO Lisa Su Says Concerns About an AI Bubble Are Overblown

#2 out of 21

technology1h ago

AMD CEO Lisa Su Says Concerns About an AI Bubble Are Overblown

Wired.com

Su dismisses concerns about an AI bubble, calling the risk 'somewhat overstated' and not the primary worry for the AI sector.
She argues AI growth depends on more computing power and data centers to support future models.
Su notes AMD’s leadership since 2014 has boosted its market cap from $2 billion to about $300 billion.
The CEO describes ongoing bets on AI data centers to power future AI deployments.
Su confirms AMD will ship MI308 chips to China with a 15% Trump-era tax, amid export controls.
AMD's OpenAI deal includes deploying 6 gigawatts of Instinct GPUs and a stock option arrangement.
Su says competition from Nvidia and others is not her primary worry; the focus is moving faster on innovation.
She sees AI as still in its infancy, requiring more computing power for future models.
The interview underscores AMD’s strategy to support AI data centers and broader AI infrastructure growth.

Vote 0

OpenAI tipped to launch ads on ChatGPT — and there's already a huge backlash

#3 out of 2199.84%

technology4h ago

OpenAI tipped to launch ads on ChatGPT — and there's already a huge backlash

Tomsguide.com

The piece discusses a potential move to add ads or app suggestions inside ChatGPT amid growing backlash.
OpenAI says there is no financial component behind the shown suggestions, defending them as not paid ads.
User backlash centers on the intrusion and potential shift in ChatGPT's trusted role as a personal assistant.
The article notes OpenAI's ongoing competition with Gemini 3.0 as context for monetization discussions.
tom's Guide frames the discussion around whether such features would be clearly labeled as ads or paid placements.
The piece highlights the broader question of preserving user trust as AI tools become more monetized.
The article mentions that there is no confirmed advertising platform yet, keeping expectations uncertain.
The coverage includes quotes from OpenAI officials and reactions from users on social media.
Tom's Guide also discusses potential future implications if monetization progresses in ChatGPT.
The article emphasizes the ongoing debate over whether paid features could undermine the core value of the tool.

Vote 0

ChatGPT 5.1 vs. Claude Opus 4.5: ChatGPT Has More Features, but Talks in Lists

#4 out of 21

technology1d ago

ChatGPT 5.1 vs. Claude Opus 4.5: ChatGPT Has More Features, but Talks in Lists

Cnet.com

ChatGPT 5.1 is described as more feature-rich and offers better integration of images and shopping links within chats.
Claude Opus 4.5 is noted for tighter, more natural prose and concise long-form responses.
In testing, Claude is described as sounding more human and less robotic than ChatGPT in many scenarios.
The article notes ChatGPT's voice mode and visual AI work well, contrasting with Claude's more limited capabilities in those areas.
The piece mentions pricing differences, noting Claude can be cheaper with an annual plan but may have occasional server issues.
The article frames the overall takeaway as a user-dependent choice between research depth and conversational style.
The article highlights ChatGPT's advantage in generating images and broader web integration.
The piece notes Claude's stronger performance in coding tasks within the chat window.

Vote 0

OpenAI’s privacy defense didn’t hold up in court – and that could shape how your AI chat data is treated going forward

#5 out of 21

technology7h ago

OpenAI’s privacy defense didn’t hold up in court – and that could shape how your AI chat data is treated going forward

Techradar.com

A U.S. magistrate orders OpenAI to hand over up to 20 million ChatGPT logs to prosecutors for copyright review.
OpenAI argues the disclosure risks user privacy despite safeguards described in the ruling.
The logs will be de-identified before transfer, with historical data covered by the ruling potentially included.
The case centers on alleged use of Times content to train AI, prompting the Times to seek user logs for rebuttal.
OpenAI plans to appeal the ruling, citing privacy commitments to users.
The ruling could foreshadow future privacy battles as courts scrutinize AI training data.
The court notes multiple privacy protections exist in this discovery process.
OpenAI’s data hosting choices are at issue as the case progresses.
This marks OpenAI’s first court-ordered sharing of consumer ChatGPT data.
The ruling may influence how providers balance privacy and legal obligations going forward.

Vote 0

OpenAI to acquire Neptune, a startup that helps with AI model training

#6 out of 217.9K est. views

technology9h ago

OpenAI to acquire Neptune, a startup that helps with AI model training

Cnbc.com and 1 more

OpenAI is pursuing an acquisition of Neptune, a startup focused on AI model training and related infrastructure to strengthen its development capabilities.
The strategic move aims to boost OpenAI's ability to train larger, more complex models by integrating Neptune's tools into its workflow.
Neptune provides tools and infrastructure that support machine learning workloads, aligning with OpenAI’s training needs.
The initial release did not disclose financial terms or integration timelines for the deal.
Analysts view the acquisition as a strategic step to broaden OpenAI's training efficiency and support for enterprise-oriented AI solutions.
The announcement signals OpenAI's continued push to scale and optimize the development of advanced AI systems across settings.
Neptune’s role as an AI model-training startup contributes new capabilities to OpenAI’s ecosystem.
Market reaction to the deal remained unclear due to the absence of disclosed terms at the time of the release.
The acquisition reflects a broader industry trend of consolidating AI training resources to accelerate development timelines.

Vote 1

I gave five major chatbots a piece of writing to edit — here’s the one that nailed the task

#7 out of 21

technology12h ago

I gave five major chatbots a piece of writing to edit — here’s the one that nailed the task

Tomsguide.com

The test gave each chatbot a 600–700 word article and asked for grammatical errors to be found and rewritten.
ChatGPT delivered extensive edits and explained changes, though some edits were not essential.
Gemini offered detailed error grouping and a full rewritten version, going beyond basic corrections.
Copilot provided brief corrections and a rewritten example, prioritizing end results over detailed reasoning.
Grok produced a long response with many issues but also useful suggestions on wordiness and repetition.
The study identified ChatGPT and Gemini as the strongest performers overall in the editor task.
Executive takeaway: editors benefit from a balance of detailed edits and reasoning to guide future writing.
The test emphasized the importance of user guidance and explanation alongside edits.
The article notes that responses varied in depth, with some editors offering extensive rationale and others brief edits.
The comparison includes how editors handle tone and style adjustments in addition to grammar fixes.

Vote 0

OpenAI awards millions to nonprofits under new foundation structure

#8 out of 21

technology12h ago

OpenAI awards millions to nonprofits under new foundation structure

Euronews.com and 1 more

OpenAI Foundation awarded $40.5 million to over 200 nonprofits in the first grants round.
Grants total the foundation’s health research and AI resilience funding commitments.
Awards are unrestricted, letting nonprofits use funds as they choose.
Applications were reviewed by outside advisers with board finalizing decisions.
Nonprofits with annual budgets between $500K and $10M were eligible.
Three Rivers Young People’s Orchestras in Pittsburgh received $90,000.
The organization plans AI-assisted rehearsal notes and clip archiving tools.
OpenAI rebranded its nonprofit as the OpenAI Foundation in October.
A new call for applications for AI and mental health research was issued by the for-profit wing.
The donations follow OpenAI’s broader philanthropic efforts and governance changes.

Vote 1

OpenAI's chief researcher says Mark Zuckerberg 'hand-delivered soup' to an employee in a recruiting effort

#9 out of 2138.8K est. views

technology16h ago

OpenAI's chief researcher says Mark Zuckerberg 'hand-delivered soup' to an employee in a recruiting effort

Mrf.lu

Meta CEO Mark Zuckerberg reportedly hand-delivered homemade soup to an OpenAI employee as part of a recruiting effort.
OpenAI says the recruiting push has included attempts to poach top researchers and engineers.
The competitor activity underscores the AI talent race between Meta and OpenAI, with other reports of signing bonuses.
OpenAI notes that recruitment efforts focus on researchers and engineers with breakthroughs or LLM expertise.
Chen says OpenAI has been fairly successful at retaining top talent amid poaching pressure.
Meta claims to be pursuing AI expertise as OpenAI experiences ongoing talent turnover.
The story highlights the broader AI talent wars among tech giants like Meta and OpenAI.
Business Insider notes that the OpenAI recruitment tension has been rising for years.
The reporting emphasizes that talent movement shapes the AI development timeline and leadership edge.
The piece notes the broader context of competitions over top AI researchers and engineers.

Vote 0

#10

Beware of these scam apps on the Mac Store

#10 out of 21

technology23h ago

Beware of these scam apps on the Mac Store

Techradar.com

Latest: Mac App Store again hosts apps mimicking Google Gemini and OpenAI's ChatGPT, signaling ongoing brand impersonation.
Neural Techlabs repeatedly published impersonating apps, with new iterations despite removals.
Removals have occurred, but new versions keep appearing, revealing flaws in Apple's review process.
Users may unknowingly download apps that access sensitive information, increasing security risks.
Investigations suggest multiple developer accounts may belong to the same group, enhancing deception.
Examples include ‘AI Chat Bot for Google Gemini,’ closely mirroring Google's branding.
Past removals include ‘AI Chat Bot Ask Assistant,’ indicating recurring violations of Apple’s rules.
Brand impersonation can bypass basic security measures and risk device integrity.
Apple’s review process is under renewed scrutiny as impersonators persist.
Security-conscious users should verify developer credentials and remain wary of brand impersonation.

Vote 0

Explore Your Interests

Unlimited Access

Personalized Feed

Full Experience

By continuing, you agree to the Privacy Policy.. You also agree to receive our newsletters, you can opt-out any time.

Explore Your Interests

Create an account and enjoy content that interests you with your personalized feed

Unlimited Access

Personalized Feed

Full Experience

By continuing, you agree to the Privacy Policy.. You also agree to receive our newsletters, you can opt-out any time.