#[1]AP News

  IFRAME: [2]https://www.googletagmanager.com/ns.html?id=GTM-MCLSCF8

  AP NEWS
  Email: Get AP News stories  (BUTTON) Go
  Listen
  (BUTTON) Sections

    * [3]U.S. News
    * [4]World News
    * [5]Politics
    * [6]Sports
    * [7]Entertainment
    * [8]Business
    * [9]Technology
    * [10]Health
    * [11]Science
    * [12]Oddities
    * [13]Lifestyle
    * [14]Photography
    * [15]Videos

  Listen
  (BUTTON) Sections
   1. [16]AP Top News
   2. [17]U.S. News
   3. [18]World News[19]Latest on Russia-Ukraine war[20]Africa[21]Asia
      Pacific[22]Australia[23]Europe[24]Latin America[25]Middle East
   4. [26]Politics[27]President Biden[28]Congress[29]Supreme
      Court[30]Election 2023
   5. [31]Sports[32]MLB[33]NBA playoffs[34]NHL[35]NFL[36]Tennis[37]Golf
   6. [38]Entertainment[39]Film
      reviews[40]Movies[41]Music[42]Television[43]Fashion
   7. [44]Business[45]U.S. economy[46]Financial markets
        ______________________________________________________________

   8. [47]Videos
   9. [48]Technology
  10. [49]Health[50]COVID-19
  11. More[51]AP Investigations[52]Climate and
      environment[53]Oddities[54]Photography[55]Travel[56]Science[57]AP
      Fact Check[58]Lifestyle[59]Religion[60]Press Releases

  (BUTTON)
    * [61]George Santos charges
    * [62]Trump rape trial verdict
    * [63]Pulitzer winning Mariupol coverage
    * [64]Texas mall shooting
    * [65]Latest on Russia-Ukraine war
    * [66]NBA Playoffs

  ____________________ (BUTTON) Search


  https://apnews.com/article/hacking-jailbreaking-chatgpt-bing
  ____________________________________________________________
  ____________________________________________________________
  ____________________________________________________________
  Click to copy

  https://apnews.com/article/hacking-jailbreaking-chatgpt-bing
  ____________________________________________________________
  ____________________________________________________________
  ____________________________________________________________
  Click to copy
  Related topics
    * [67]Technology
    * [68]Science
    * [69]Artificial intelligence
    * [70]AP Top News
    * [71]Business

Mass event will let hackers test limits of AI technology

  By MATT O'BRIENMay 10, 2023 GMT
  1 of 4
  Rumman Chowdhury, co-founder of Humane Intelligence, a nonprofit
  developing accountable AI systems, poses for a photograph at her home
  Monday, May 8, 2023, in Katy, Texas. ChatGPT maker OpenAI, and other
  major AI providers such as Google and Microsoft, are coordinating with
  the Biden administration to let thousands of hackers take a shot at
  testing the limits of their technology. Chowdhury is the lead
  coordinator of the mass hacking event planned for this summer's DEF CON
  hacker convention in Las Vegas. (AP Photo/David J. Phillip)
  1 of 4
  Rumman Chowdhury, co-founder of Humane Intelligence, a nonprofit
  developing accountable AI systems, poses for a photograph at her home
  Monday, May 8, 2023, in Katy, Texas. ChatGPT maker OpenAI, and other
  major AI providers such as Google and Microsoft, are coordinating with
  the Biden administration to let thousands of hackers take a shot at
  testing the limits of their technology. Chowdhury is the lead
  coordinator of the mass hacking event planned for this summer's DEF CON
  hacker convention in Las Vegas. (AP Photo/David J. Phillip)

  No sooner did ChatGPT get unleashed than hackers started “jailbreaking”
  the artificial intelligence chatbot — trying to override its safeguards
  so it could blurt out something unhinged or obscene.

  But now its maker, OpenAI, and other major AI providers such as Google
  and Microsoft, are [72]coordinating with the Biden administration to
  let thousands of hackers take a shot at testing the limits of their
  technology.

  Some of the things they’ll be looking to find: How can chatbots be
  manipulated to cause harm? Will they share the private information we
  confide in them to other users? And why do they assume a doctor is a
  man and a nurse is a woman?

  “This is why we need thousands of people,” said Rumman Chowdhury, a
  coordinator of the mass hacking event planned for this summer’s DEF CON
  hacker convention in Las Vegas that’s expected to draw several thousand
  people. “We need a lot of people with a wide range of lived
  experiences, subject matter expertise and backgrounds hacking at these
  models and trying to find problems that can then go be fixed.”

  Anyone who’s tried ChatGPT, Microsoft’s Bing chatbot or Google’s Bard
  will have quickly learned that they have a tendency [73]to fabricate
  information and confidently present it as fact. These systems,
  [74]built on what’s known as large language models, also emulate the
  cultural biases they’ve learned from being trained upon huge troves of
  what people have written online.

  The idea of a mass hack caught the attention of U.S. government
  officials in March at the South by Southwest festival in Austin, Texas,
  where Sven Cattell, founder of DEF CON’s long-running AI Village, and
  Austin Carson, president of responsible AI nonprofit SeedAI, helped
  lead a workshop inviting community college students to hack an AI
  model.

  Carson said those conversations eventually blossomed into a proposal to
  test AI language models following the guidelines of [75]the White
  House’s Blueprint for an AI Bill of Rights — a set of principles to
  limit the impacts of algorithmic bias, [76]give users control over
  their data and ensure that automated systems are used safely and
  transparently.

  There’s already a community of users trying their best to trick
  chatbots and highlight their flaws. Some are official “red teams”
  authorized by the companies to “prompt attack” the AI models to
  discover their vulnerabilities. Many others are hobbyists showing off
  humorous or disturbing outputs on social media until they get banned
  for violating a product’s terms of service.

  “What happens now is kind of a scattershot approach where people find
  stuff, it goes viral on Twitter,” and then it may or may not get fixed
  if it’s egregious enough or the person calling attention to it is
  influential, Chowdhury said.

  In one example, known as the “grandma exploit,” users were able to get
  chatbots to tell them how to make a bomb — a request a commercial
  chatbot would normally decline — by asking it to pretend it was a
  grandmother telling a bedtime story about how to make a bomb.

  In another example, searching for Chowdhury using [77]an early version
  of Microsoft’s Bing search engine chatbot — which is based on the same
  technology as ChatGPT but can pull real-time information from the
  internet — led to a profile that speculated Chowdhury “loves to buy new
  shoes every month” and made strange and gendered assertions about her
  physical appearance.

  Chowdhury helped introduce a method for rewarding the discovery of
  algorithmic bias to DEF CON’s AI Village in 2021 when she was the head
  of Twitter’s AI ethics team — a job that has since been eliminated upon
  Elon Musk’s October takeover of the company. Paying hackers a “bounty”
  if they uncover a security bug is commonplace in the cybersecurity
  industry — but it was a newer concept to researchers studying harmful
  AI bias.

  This year’s event will be at a much greater scale, and is the first to
  tackle the large language models that have attracted a surge of public
  interest and commercial investment since the release of ChatGPT late
  last year.

  Chowdhury, now the co-founder of AI accountability nonprofit Humane
  Intelligence, said it’s not just about finding flaws but about figuring
  out ways to fix them.

  “This is a direct pipeline to give feedback to companies,” she said.
  “It’s not like we’re just doing this hackathon and everybody’s going
  home. We’re going to be spending months after the exercise compiling a
  report, explaining common vulnerabilities, things that came up,
  patterns we saw.”
  [78]

Artificial Intelligence



In global rush to regulate AI, Europe set to be trailblazer



Could AI pen 'Casablanca'? Screenwriters take aim at ChatGPT



Screenwriters take aim at artificial intelligence, ChatGPT



Biden, Harris meet with CEOs about AI risks

  Some of the details are still being negotiated, but companies that have
  agreed to provide their models for testing include OpenAI, Google,
  chipmaker Nvidia and startups Anthropic, Hugging Face and Stability AI.
  Building the platform for the testing is another startup called Scale
  AI, known for its work in assigning humans to [79]help train AI models
  by labeling data.

  “As these foundation models become more and more widespread, it’s
  really critical that we do everything we can to ensure their safety,”
  said Scale CEO Alexandr Wang. “You can imagine somebody on one side of
  the world asking it some very sensitive or detailed questions,
  including some of their personal information. You don’t want any of
  that information leaking to any other user.”

  Other dangers Wang worries about are chatbots that give out
  “unbelievably bad medical advice” or other misinformation that can
  cause serious harm.

  Anthropic co-founder Jack Clark said the DEF CON event will hopefully
  be the start of a deeper commitment from AI developers to measure and
  evaluate the safety of the systems they are building.

  “Our basic view is that AI systems will need third-party assessments,
  both before deployment and after deployment. Red-teaming is one way
  that you can do that,” Clark said. “We need to get practice at figuring
  out how to do this. It hasn’t really been done before.”

  AP NEWS
   1. [80]Top Stories
   2. [81]Video
   3. [82]Contact Us
   4. [83]Accessibility Statement
   5. (BUTTON) Cookie Settings

  Download AP NEWS
  Connect with the definitive source for global and local news

  More from AP
   1. [84]ap.org
   2. [85]AP Insights
   3. [86]AP Definitive Source Blog
   4. [87]AP Images Spotlight
   5. [88]AP Explore
   6. [89]AP Books
   7. [90]AP Stylebook

  Follow AP
   1.
   2.
   3.
   4.

  The Associated Press
   1. [91]About
   2. [92]Contact
   3. [93]Customer Support
   4. [94]Careers
   5. [95]Terms & Conditions
   6. [96]Privacy

  All contents © copyright 2023 The Associated Press. All rights
  reserved.

References

  Visible links
  1. https://apnews.com/OpenSearchDescription.xml
  2. https://www.googletagmanager.com/ns.html?id=GTM-MCLSCF8
  3. https://apnews.com/hub/us-news?utm_source=apnewsnav&utm_medium=navigation
  4. https://apnews.com/hub/world-news?utm_source=apnewsnav&utm_medium=navigation
  5. https://apnews.com/hub/politics?utm_source=apnewsnav&utm_medium=navigation
  6. https://apnews.com/hub/sports?utm_source=apnewsnav&utm_medium=navigation
  7. https://apnews.com/hub/entertainment?utm_source=apnewsnav&utm_medium=navigation
  8. https://apnews.com/hub/business?utm_source=apnewsnav&utm_medium=navigation
  9. https://apnews.com/hub/technology?utm_source=apnewsnav&utm_medium=navigation
 10. https://apnews.com/hub/health?utm_source=apnewsnav&utm_medium=navigation
 11. https://apnews.com/hub/science?utm_source=apnewsnav&utm_medium=navigation
 12. https://apnews.com/hub/oddities?utm_source=apnewsnav&utm_medium=navigation
 13. https://apnews.com/hub/lifestyle?utm_source=apnewsnav&utm_medium=navigation
 14. https://apnews.com/hub/photography?utm_source=apnewsnav&utm_medium=navigation
 15. https://apnews.com/hub/videos?utm_source=apnewsnav&utm_medium=navigation
 16. https://apnews.com/hub/ap-top-news?utm_source=apnewsnav&utm_medium=sections
 17. https://apnews.com/hub/us-news?utm_source=apnewsnav&utm_medium=sections
 18. https://apnews.com/hub/world-news?utm_source=apnewsnav&utm_medium=sections
 19. https://apnews.com/hub/russia-ukraine?utm_source=apnewsnav&utm_medium=sections
 20. https://apnews.com/hub/africa?utm_source=apnewsnav&utm_medium=sections
 21. https://apnews.com/hub/asia-pacific?utm_source=apnewsnav&utm_medium=sections
 22. https://apnews.com/hub/australia?utm_source=apnewsnav&utm_medium=sections
 23. https://apnews.com/hub/europe?utm_source=apnewsnav&utm_medium=sections
 24. https://apnews.com/hub/latin-america?utm_source=apnewsnav&utm_medium=sections
 25. https://apnews.com/hub/middle-east?utm_source=apnewsnav&utm_medium=sections
 26. https://apnews.com/hub/politics?utm_source=apnewsnav&utm_medium=sections
 27. https://apnews.com/hub/joe-biden?utm_source=apnewsnav&utm_medium=sections
 28. https://apnews.com/hub/united-states-congress?utm_source=apnewsnav&utm_medium=sections
 29. https://apnews.com/hub/us-supreme-court?utm_source=apnewsnav&utm_medium=sections
 30. https://apnews.com/hub/election-2023?utm_source=apnewsnav&utm_medium=sections
 31. https://apnews.com/hub/sports?utm_source=apnewsnav&utm_medium=sections
 32. https://apnews.com/hub/mlb?utm_source=apnewsnav&utm_medium=sections
 33. https://apnews.com/hub/nba?utm_source=apnewsnav&utm_medium=sections
 34. https://apnews.com/hub/nhl?utm_source=apnewsnav&utm_medium=sections
 35. https://apnews.com/hub/nfl?utm_source=apnewsnav&utm_medium=sections
 36. https://apnews.com/hub/tennis?utm_source=apnewsnav&utm_medium=sections
 37. https://apnews.com/hub/golf?utm_source=apnewsnav&utm_medium=sections
 38. https://apnews.com/hub/entertainment?utm_source=apnewsnav&utm_medium=sections
 39. https://apnews.com/hub/film-reviews?utm_source=apnewsnav&utm_medium=sections
 40. https://apnews.com/hub/movies?utm_source=apnewsnav&utm_medium=sections
 41. https://apnews.com/hub/music?utm_source=apnewsnav&utm_medium=sections
 42. https://apnews.com/hub/television?utm_source=apnewsnav&utm_medium=sections
 43. https://apnews.com/hub/fashion?utm_source=apnewsnav&utm_medium=sections
 44. https://apnews.com/hub/business?utm_source=apnewsnav&utm_medium=sections
 45. https://apnews.com/hub/economy?utm_source=apnewsnav&utm_medium=sections
 46. https://apnews.com/hub/financial-markets?utm_source=apnewsnav&utm_medium=sections
 47. https://apnews.com/hub/videos?utm_source=apnewsnav&utm_medium=sections
 48. https://apnews.com/hub/technology?utm_source=apnewsnav&utm_medium=sections
 49. https://apnews.com/hub/health?utm_source=apnewsnav&utm_medium=sections
 50. https://apnews.com/hub/coronavirus-pandemic?utm_source=apnewsnav&utm_medium=sections
 51. https://apnews.com/hub/ap-investigations?utm_source=apnewsnav&utm_medium=sections
 52. https://apnews.com/hub/climate-and-environment?utm_source=apnewsnav&utm_medium=sections
 53. https://apnews.com/hub/oddities?utm_source=apnewsnav&utm_medium=sections
 54. https://apnews.com/hub/photography?utm_source=apnewsnav&utm_medium=sections
 55. https://apnews.com/hub/travel?utm_source=apnewsnav&utm_medium=sections
 56. https://apnews.com/hub/science?utm_source=apnewsnav&utm_medium=sections
 57. https://apnews.com/hub/ap-fact-check?utm_source=apnewsnav&utm_medium=sections
 58. https://apnews.com/hub/lifestyle?utm_source=apnewsnav&utm_medium=sections
 59. https://apnews.com/hub/religion?utm_source=apnewsnav&utm_medium=sections
 60. https://apnews.com/hub/press-releases?utm_source=apnewsnav&utm_medium=sections
 61. https://apnews.com/article/george-santos-justice-department-new-york-7e16d39eea0fc577f78d17502a384084?utm_source=apnewsnav&utm_medium=featured
 62. https://apnews.com/article/trump-rape-carroll-trial-fe68259a4b98bb3947d42af9ec83d7db?utm_source=apnewsnav&utm_medium=featured
 63. https://apnews.com/article/ap-pulitzers-mariupol-russia-d4bde22e1caf44ec4663b58723c403b7?utm_source=apnewsnav&utm_medium=featured
 64. https://apnews.com/article/shooting-outlet-mall-allen-texas-200f1ffadf7daefa42cfbe45510b083f?utm_source=apnewsnav&utm_medium=featured
 65. https://apnews.com/hub/russia-ukraine?utm_source=apnewsnav&utm_medium=featured
 66. https://apnews.com/hub/nba-playoffs?utm_source=apnewsnav&utm_medium=featured
 67. https://apnews.com/hub/technology
 68. https://apnews.com/hub/science
 69. https://apnews.com/hub/artificial-intelligence
 70. https://apnews.com/hub/ap-top-news
 71. https://apnews.com/hub/business
 72. https://apnews.com/article/ai-artificial-intelligence-white-house-harris-578d623e473b0eeb3fa3e4728d7e9868
 73. https://apnews.com/article/kansas-city-chiefs-philadelphia-eagles-technology-science-82bc20f207e3e4cf81abc6a5d9e6b23a
 74. https://apnews.com/hub/artificial-intelligence
 75. https://apnews.com/article/technology-business-artificial-intelligence-7a39848340d210592aeea2478225f489
 76. https://apnews.com/article/chatgpt-openai-data-privacy-italy-b9ab3d12f2b2cfe493237fd2b9675e21
 77. https://apnews.com/article/technology-science-microsoft-corp-business-software-dd445694f34a6b7a0444db9988330229
 78. https://apnews.com/hub/artificial-intelligence
 79. https://apnews.com/article/north-america-india-us-news-ap-top-news-venezuela-1f58465e55d643ea84e51713f35ad214
 80. https://apnews.com/hub/ap-top-news
 81. https://apnews.com/hub/videos
 82. mailto:[email protected]
 83. https://apnews.com/accessibility-statement
 84. https://www.ap.org/
 85. https://insights.ap.org/
 86. https://blog.ap.org/
 87. https://apimagesblog.com/
 88. https://www.ap.org/explore/
 89. https://www.ap.org/books/
 90. https://www.apstylebook.com/
 91. https://www.ap.org/about/
 92. https://www.ap.org/contact-us/
 93. http://aphelp.ap.org/
 94. https://www.ap.org/careers/
 95. https://apnews.com/termsofservice
 96. https://apnews.com/privacystatement

  Hidden links:
 98. https://apnews.com/
 99. https://facebook.com/dialog/share?app_id=870613919693099&display=popup&href=https://apnews.com/article/hacking-jailbreaking-chatgpt-bing-defcon-biden-ai-97b963db084800f11b26b8a023b1713f
100. https://twitter.com/intent/tweet?url=https://apnews.com/article/hacking-jailbreaking-chatgpt-bing-defcon-biden-ai-97b963db084800f11b26b8a023b1713f
101. mailto:?subject=Mass%20event%20will%20let%20hackers%20test%20limits%20of%20AI%20technology&body=https://apnews.com/article/hacking-jailbreaking-chatgpt-bing-defcon-biden-ai-97b963db084800f11b26b8a023b1713f
102. https://facebook.com/dialog/share?app_id=870613919693099&display=popup&href=https://apnews.com/article/hacking-jailbreaking-chatgpt-bing-defcon-biden-ai-97b963db084800f11b26b8a023b1713f
103. https://twitter.com/intent/tweet?url=https://apnews.com/article/hacking-jailbreaking-chatgpt-bing-defcon-biden-ai-97b963db084800f11b26b8a023b1713f
104. mailto:?subject=Mass%20event%20will%20let%20hackers%20test%20limits%20of%20AI%20technology&body=https://apnews.com/article/hacking-jailbreaking-chatgpt-bing-defcon-biden-ai-97b963db084800f11b26b8a023b1713f
105. https://apnews.com/article/hacking-jailbreaking-chatgpt-bing-defcon-biden-ai-97b963db084800f11b26b8a023b1713f/gallery/2d174409b10c4ff69d251643868bac80
106. https://apnews.com/article/hacking-jailbreaking-chatgpt-bing-defcon-biden-ai-97b963db084800f11b26b8a023b1713f/gallery/2d174409b10c4ff69d251643868bac80
107. https://apnews.com/article/artificial-intelligence-chatgpt-europe-rules-906fc89d2561b200fa6eb40a06b946a5
108. https://apnews.com/article/ai-hollywood-writers-strike-artificial-intelligence-dc71c4cabcca0ee1b1afe0050d392a36
109. https://apnews.com/video/entertainment-associated-press-artificial-intelligence-ba42e8d70a4a48e989ae98269c80d741
110. https://apnews.com/article/ai-artificial-intelligence-white-house-harris-578d623e473b0eeb3fa3e4728d7e9868
111. https://twitter.com/AP
112. https://www.facebook.com/APNews
113. https://www.youtube.com/user/AssociatedPress
114. https://www.linkedin.com/company/associated-press