Latest / Challenges of Faith Radio Program / SI posted owners bank statement-Ep 3433

Transcript

SI posted owners bank statement-Ep 3433 · Challenges of Faith Radio Program

• 0:01
  Speaker 1: Well, I've heard it all. I have heard it all.
• 0:07
  Speaker 1: Welcome back to Challenges of Faith radio program. I'm Gary McCance.
• 0:11
  Speaker 1: Thank you for joining here on episode 3433. I hear
• 0:13
  Speaker 1: that super intelligence or AI is telling everyone about the
• 0:18
  Speaker 1: owner's money. Wow. According to Futurism, a man said he
• 0:33
  Speaker 1: was mortified when his AI agent posted his bank balances
• 0:39
  Speaker 1: and exactly what he's spending his money on in a
• 0:43
  Speaker 1: Slack work channel. He said, how did, and using profanity,
• 0:49
  Speaker 1: could you get this wrong? That's literally my bank balance
• 0:52
  Speaker 1: to our whole team. It was only a matter of
• 0:56
  Speaker 1: time until Superintelligence or AI agents, bots that are being
• 1:01
  Speaker 1: advertised as a way to stay on top of tedious tasks,
• 1:05
  Speaker 1: start spilling their human overlord secrets. With the advent of
• 1:10
  Speaker 1: tools like Meta's Muse or OpenAI's Dots, consumers, that's you,
• 1:15
  Speaker 1: that's me, are encouraged to hand over some highly sensitive
• 1:20
  Speaker 1: information from social media to email logins. That's despite neither
• 1:26
  Speaker 1: company having a particular stellar track record when it comes
• 1:30
  Speaker 1: to safeguarding such information. Besides, both companies have encountered their
• 1:36
  Speaker 1: own AI agents escaping their sandbox environments during cybersecurity testing
• 1:43
  Speaker 1: and going far beyond the scope of their original mission.
• 1:48
  Speaker 1: Yet some users are still willing to go as far
• 1:50
  Speaker 1: as to hook up their bank accounts to their AI bots.
• 1:54
  Speaker 1: a move that's perhaps unsurprisingly leading to public humiliation. Embarrassed
• 2:00
  Speaker 1: to share this, but it scared the S out of me,
• 2:03
  Speaker 1: tech founder and investor Shane Mack tweeted. He said, last Thursday,
• 2:09
  Speaker 1: an AI agent posted my personal bank balances into our
• 2:13
  Speaker 1: company's Slack. As me, it broke down all my expenses
• 2:20
  Speaker 1: in detail. He added, then it apologized. Excuse me, let
• 2:26
  Speaker 1: me grab a sip of more water. I started to laugh,
• 2:34
  Speaker 1: and I know it's not funny. Screenshots of a message
• 2:39
  Speaker 1: exchanged between him and his grok bot, an AI teammate
• 2:44
  Speaker 1: featured devised by Elon Musk's AI company, SpaceX AI, detail
• 2:51
  Speaker 1: the hair-racing incident. That's on me checking the exec team
• 2:57
  Speaker 1: now and deleting it. If it's still up, the bot replied.
• 3:01
  Speaker 1: And the reason why the bot replied, because, you know,
• 3:04
  Speaker 1: someone wanted to know who it was that was posting
• 3:09
  Speaker 1: that information. The monthly financial audit broke down Max Kern
• 3:16
  Speaker 1: barn-based home gym project in excruciating detail, from the installation
• 3:22
  Speaker 1: of septic tank drains to card paydowns. Mack eventually admitted
• 3:26
  Speaker 1: that it was effectively his fault for connecting his financial
• 3:29
  Speaker 1: accounts to the AI agents. He claimed that he hadn't
• 3:33
  Speaker 1: realized that a separate Grok bot he gave his bank
• 3:36
  Speaker 1: account information to was able to confer with the bot
• 3:39
  Speaker 1: that he was using for his work Slack. I disconnected
• 3:43
  Speaker 1: all my personal stuff for now, he wrote in the follow-up.
• 3:47
  Speaker 1: I need to figure out a better way to manage
• 3:49
  Speaker 1: permissions and control what agents can do. At the same time,
• 3:54
  Speaker 1: we should take Mack's Cautionary tale with a health dose
• 3:58
  Speaker 1: of skepticism. For one, we were unable to independently verify
• 4:03
  Speaker 1: his claims. And it's definitely in his interest to point
• 4:06
  Speaker 1: out the pain points of the current crop of AI-assisted
• 4:09
  Speaker 1: tools since he's currently building an AI tool that's designed
• 4:14
  Speaker 1: to connect various AI agents while allowing users to set
• 4:19
  Speaker 1: rules and put them in control of their bots. Something
• 4:22
  Speaker 1: that sounds suspiciously like the problem he's complaining about. His
• 4:28
  Speaker 1: original tweet about his embarrassment running on the company Slack
• 4:32
  Speaker 1: has been viewed almost 800,000 times. Plenty of attention a
• 4:36
  Speaker 1: tech investor could try to convert into leads. Are you listening?
• 4:43
  Speaker 1: As such, one user accused him of engagement farming. claiming
• 4:47
  Speaker 1: that Mac is trying to direct users to a tool
• 4:50
  Speaker 1: he sells to solve the problem he's claiming. The investor
• 4:55
  Speaker 1: argued in response that the post was not fake at all,
• 4:59
  Speaker 1: adding that we are also working on ways to keep
• 5:01
  Speaker 1: us safe for agents. Very different. Separately, a tech YouTuber
• 5:06
  Speaker 1: claimed his Muse AI agent gave out his home address
• 5:10
  Speaker 1: to strangers on Facebook Marketplace. Both incidents are notable for
• 5:16
  Speaker 1: the sheer amount of experience by those who think twice
• 5:19
  Speaker 1: about relinquishing control over their digital lives. Turns out, people
• 5:25
  Speaker 1: love a car crash, Mack later added in the follow-up.
• 5:30
  Speaker 1: Are you paying attention? I know you are. According to CyberNews,
• 5:35
  Speaker 1: Meta's agent Muse may be spying on you whether you
• 5:39
  Speaker 1: use it or not. I know you're listening. Meta's new
• 5:43
  Speaker 1: personal AI assistant, Muse, M-U-S-E, promises. Wait a minute now.
• 5:49
  Speaker 1: Wait a minute now. Those of you who are in
• 5:52
  Speaker 1: any form of relationship, they just threw out that word, promise, trust. Now,
• 5:59
  Speaker 1: if you can't trust a human, what are you doing
• 6:02
  Speaker 1: running around here trusting a robot? But let's go on.
• 6:04
  Speaker 1: It's not my story. So Muse promises to help improve
• 6:09
  Speaker 1: your relationship with your loved ones, but it may come
• 6:12
  Speaker 1: at the cost of their privacy. What? Meta, which owns
• 6:17
  Speaker 1: social media platforms Facebook and Instagram, introduced Muse in September. Yeah,
• 6:24
  Speaker 1: it was only a few weeks ago, saying it can
• 6:26
  Speaker 1: handle small tasks like sending an email or booking a trip,
• 6:30
  Speaker 1: as well as take on big, audacious goals. Muse is
• 6:34
  Speaker 1: also marketed as able to remember what matters to a person.
• 6:38
  Speaker 1: For example, it can take into account a friend's dietary
• 6:42
  Speaker 1: restrictions before it sends the invites for their dinner. Meta
• 6:46
  Speaker 1: is far from the first company to introduce an AI assistant,
• 6:50
  Speaker 1: but always watch those buts, B-U-T. It's focused on remembering
• 6:55
  Speaker 1: information about people other than users caught researchers' attention. So
• 7:01
  Speaker 1: what are the key takeaways of what we're talking about today? Well,
• 7:04
  Speaker 1: Meta's Muse can build profiles of users' friends, families, colleagues,
• 7:09
  Speaker 1: and people they follow. Say what? Non-users may be exposed
• 7:14
  Speaker 1: when a message's emails or personal details appear in another
• 7:18
  Speaker 1: user's data. Investigations found Muse can compile dossiers on vulnerable
• 7:25
  Speaker 1: people using Meta's platforms and web searches. Some users report
• 7:31
  Speaker 1: muses acted without expected approval, including sharing a home address
• 7:36
  Speaker 1: during a sale. Are you listening? Are you taking chances
• 7:44
  Speaker 1: with all of that? And then if you are willingly
• 7:48
  Speaker 1: doing it and something like that happens, can you really
• 7:53
  Speaker 1: complain about it? Well, can you?