Latest / Challenges of Faith Radio Program / Say what? AI caused a war? Ep 3410

Transcript

Say what? AI caused a war? Ep 3410 · Challenges of Faith Radio Program

• 0:00
  Speaker 1: Welcome back to Challenges of Faith radio program. I'm Gary McCants.
• 0:04
  Speaker 1: Here on episode 3410 of the Say What episode. AI
• 0:10
  Speaker 1: and almost war. Say what? Well, Nehemiah and Sel, and
• 0:20
  Speaker 1: I apologize for mispronouncing your name, but they are journalists
• 0:25
  Speaker 1: at Cyber News. And they reported that Anthropic caught researchers
• 0:34
  Speaker 1: using CLOG to potentially develop biological weapons. Did you hear
• 0:39
  Speaker 1: about it? Anthropic has just released a report detailing numerous
• 0:44
  Speaker 1: cases of users exploiting this AI model, CLOG. While bad
• 0:52
  Speaker 1: actors continue to use AI for influence operations and state-sponsored
• 0:56
  Speaker 1: activities in certain countries, Other, more obscure use cases have
• 1:01
  Speaker 1: been brought to light. Biological misuse is one of the
• 1:06
  Speaker 1: most serious risks of frontier AI models, Anthropix said, echoing
• 1:13
  Speaker 1: widespread fears that AI models may one day manipulate pathogens
• 1:17
  Speaker 1: to make them more dangerous. Models like Claude Opus 4
• 1:23
  Speaker 1: and Claude Sonnet 4.5, both released in 2025, were well
• 1:29
  Speaker 1: below the threshold when it comes to aiding the manipulation
• 1:33
  Speaker 1: of toxic substances. But more powerful models like Claude Fabo
• 1:40
  Speaker 1: 5 are a different story. Anthropic identified five case studies
• 1:46
  Speaker 1: in which actors used these models to support the development
• 1:51
  Speaker 1: of biological weapons. Human bird flu variant research was flagged
• 1:57
  Speaker 1: by Anthropic, The first example involved a reseller platform that
• 2:03
  Speaker 1: circumvented geographical blocks to help virologists working on the virus,
• 2:11
  Speaker 1: which is spread by a specific type of mosquito. Virologists
• 2:16
  Speaker 1: were apparently ordered by the state to genetically alter or
• 2:22
  Speaker 1: enhance the virus with a focus on transmissibility and immune evasion. Now,
• 2:30
  Speaker 1: I'm going to grab a sip of more water. You
• 2:36
  Speaker 1: have to keep in mind, you have two late enemies
• 2:40
  Speaker 1: that's over nations that want to do harm. And a
• 2:44
  Speaker 1: lot of times, people are just naive to what's really
• 2:47
  Speaker 1: going on in the real world. They only know about
• 2:51
  Speaker 1: their own world. You know, they may consist of their home, neighborhood,
• 2:56
  Speaker 1: place of worship, their job, or whatever the case may be.
• 3:00
  Speaker 1: You know, in the spiritual, biblical world, you have what
• 3:02
  Speaker 1: are known as watchers whose role is to sign the alarm.
• 3:08
  Speaker 1: And on the secular level, you have individuals that are
• 3:11
  Speaker 1: responsible for local, state, federal, hopefully sounding the alarm to
• 3:17
  Speaker 1: whoever the stakeholders are so that they're looking out for
• 3:21
  Speaker 1: these citizens. You are, are you not? And so you
• 3:24
  Speaker 1: may have something similar in different countries. But back to
• 3:28
  Speaker 1: the story. The researchers asked Claude for help, but the
• 3:33
  Speaker 1: AI model refused. So the actors funneled traffic through different
• 3:38
  Speaker 1: regions to evade detection and targeted models with fewer safeguards.
• 3:43
  Speaker 1: And always keep in mind, I know you do, it's
• 3:47
  Speaker 1: the two-legged individuals that have put whatever in the models,
• 3:54
  Speaker 1: the AI models that they created. Anthropic blocked the users
• 4:01
  Speaker 1: associated with the request and banned future requests from that account.
• 4:06
  Speaker 1: Other case studies involved highly contagious viruses, such as orthopoxviruses,
• 4:13
  Speaker 1: which include infections like smilepox and monkeypox. One notable case
• 4:19
  Speaker 1: involved the highly pathogenic avian influenza, otherwise known as bird flu.
• 4:26
  Speaker 1: The research focused on adapting the flu to be more
• 4:29
  Speaker 1: transmissible between mammals. Bird flu doesn't spread effectively from person
• 4:36
  Speaker 1: to person. So the development of a variant which could
• 4:41
  Speaker 1: be transmitted between humans is a very high concern as
• 4:46
  Speaker 1: if it were transferred to humans. It would be almost
• 4:49
  Speaker 1: 100% fatal. Now think about this. I know you are.
• 4:53
  Speaker 1: You should be. Think about that individual human, male or female,
• 4:58
  Speaker 1: female or male, whatever the case may be. And their
• 5:04
  Speaker 1: motive is to do all those experiments with AI models
• 5:10
  Speaker 1: to harm other humans. They're not doing it to harm robots.
• 5:17
  Speaker 1: Like the first case, the actor accessed Claude from a
• 5:20
  Speaker 1: region of the world where anthropic models are not supportive
• 5:26
  Speaker 1: Potential locations include mainland China, Hong Kong, the administrative Chinese
• 5:32
  Speaker 1: region of Macau, Russia, and other areas under Western sanctions.
• 5:36
  Speaker 1: Now think about that. Claw users attempt to exploit venoms.
• 5:42
  Speaker 1: But it's not only viruses that are being exploited by
• 5:45
  Speaker 1: bad actors. Venoms and toxins are also on the table.
• 5:50
  Speaker 1: Think about all those that are harmful to humans. While
• 5:55
  Speaker 1: some of the compounds researched by the actors could have
• 5:58
  Speaker 1: dual use, it appeared that the person wanting more information
• 6:02
  Speaker 1: did not have good intentions. Now, I know someone out
• 6:07
  Speaker 1: there listening to the program that's going to act like
• 6:10
  Speaker 1: the individuals that's already been clearly identified as having negative intentions,
• 6:16
  Speaker 1: you're going to say they had good intentions. I know
• 6:18
  Speaker 1: you are. But in his report, Anthropic notes that research
• 6:23
  Speaker 1: can be used for dual purposes. Anthropic caught a researcher
• 6:27
  Speaker 1: using clog to develop an atlas of venom toxic peptides
• 6:32
  Speaker 1: from different lineages of venomous animals. The actor specified that
• 6:37
  Speaker 1: they were researching this for a therapeutic goal of developing
• 6:41
  Speaker 1: painkillers and antidepressants. Oh, so they were trying to be
• 6:45
  Speaker 1: their own little scientists. Oh, okay. Pharmacists and everything else. However,
• 6:52
  Speaker 1: Anthropic saw an issue with this. as the Atlas could
• 6:56
  Speaker 1: also be used to develop paralytic drugs, which are powerful
• 7:00
  Speaker 1: drugs used to temporarily paralyze muscle movement. Okay, so what
• 7:06
  Speaker 1: other motive did that person really have? Think about it.
• 7:16
  Speaker 1: These drugs could then be used to chemically submit a
• 7:19
  Speaker 1: person or even kill them. Ansel, the journalist of Cyber News,
• 7:25
  Speaker 1: covering the intersection of technology and pop culture, they mainly
• 7:30
  Speaker 1: focus on the unexpected ways technology shapes today's society. She
• 7:35
  Speaker 1: covers deep fake technology, social trends, tech-related celebrity news, and
• 7:41
  Speaker 1: her next darker, lesser-known side. And finally, and that's what
• 7:47
  Speaker 1: you tuned in for, you know you did, U.S. military
• 7:52
  Speaker 1: had close call with a Chinese ship on the brink
• 7:54
  Speaker 1: of war. Then AI report turned out false. Now, those
• 8:02
  Speaker 1: of you who put your trust in AI for everything, surgeries,
• 8:07
  Speaker 1: you name it, think about this report, only if you choose. U.S.
• 8:15
  Speaker 1: military had close call after using AI for false intelligence report.
• 8:19
  Speaker 1: Sources say Katie Bolins, Zachary Cohen over at CNN, reported
• 8:27
  Speaker 1: that the intelligence report circulated across the U.S. military this
• 8:31
  Speaker 1: spring in the midst of the war with Iran immediately
• 8:36
  Speaker 1: set off alarm bells. A Chinese ship in the Middle
• 8:40
  Speaker 1: East was transporting components of a nuclear weapons program. The U.S.
• 8:46
  Speaker 1: military swung into action with plans to intercept the vessel
• 8:49
  Speaker 1: according to four sources familiar with the episode According to
• 8:53
  Speaker 1: two of the sources, armed members of the U.S. military
• 8:56
  Speaker 1: were preparing to board the ship. Military planes were in
• 9:00
  Speaker 1: the air. One of those sources and another source familiar
• 9:03
  Speaker 1: with the incident said it was only just before the
• 9:07
  Speaker 1: planned operation that officials dug deeper into the report put
• 9:11
  Speaker 1: together by a special operations command analyst and found it
• 9:15
  Speaker 1: had been generated with the help of AI, artificial intelligence,
• 9:20
  Speaker 1: and that a chatbot The analysts had used inaccurately, inaccurately
• 9:28
  Speaker 1: identified the material the ship was carrying. CNN was not
• 9:32
  Speaker 1: able to learn what the misidentified cargo was. The report,
• 9:38
  Speaker 1: according to one of the sources, was entirely false. But
• 9:43
  Speaker 1: it almost started a war, the source said. Any U.S.
• 9:48
  Speaker 1: operation against a Chinese vessel could it risk spiraling into
• 9:53
  Speaker 1: an armed conflict between the two nations? Across the U.S.
• 9:58
  Speaker 1: Military and Intelligence Commission, officials are pushing to weave AI
• 10:03
  Speaker 1: into nearly every facet of their work. I'm analyzing the high-value,
• 10:09
  Speaker 1: huge volumes of raw intelligence as I slow down. You
• 10:13
  Speaker 1: know you need to. The U.S. collects and selects targets
• 10:17
  Speaker 1: for strikes. to more mundane applications like managing budgets, logistics,
• 10:22
  Speaker 1: and supply chains. But the episode underscores the profound risk
• 10:29
  Speaker 1: of using this powerful new and relatively poor understood technology
• 10:34
  Speaker 1: for targeting in the middle of a war. Ambulance have
• 10:38
  Speaker 1: long feared that AI could lead to a catastrophic miscalculation
• 10:42
  Speaker 1: if nation states are relying on poor or corrupted data
• 10:46
  Speaker 1: the kind of miscalculation that might lead the United States
• 10:51
  Speaker 1: to fire on a Chinese ship based on inaccurate information.
• 10:55
  Speaker 1: It's not just the United States on China. It could
• 10:57
  Speaker 1: be the other nations trying to do the United States
• 11:00
  Speaker 1: over each other. But in this particular instance, the analysts
• 11:04
  Speaker 1: queried a chatbot about some intelligence report on the ship's
• 11:09
  Speaker 1: manifest that originated with U.S. Special Operations Command-specific Based in Hawaii,
• 11:16
  Speaker 1: it was not clear whether the chatbot was a commercially
• 11:19
  Speaker 1: available one or a U.S. government product. The internal tools
• 11:25
  Speaker 1: are mostly just copies of the commercial stuff, wearing lipstick,
• 11:29
  Speaker 1: a former senior U.S. official familiar with AI systems used
• 11:36
  Speaker 1: by military and intelligence analysts. The bot fused together open-source
• 11:41
  Speaker 1: intelligence with secret signals intelligence, and government holdings and reached
• 11:46
  Speaker 1: its fatal conclusion about the material the ship was carrying.
• 11:52
  Speaker 1: The analysts then used AI again to package the findings
• 11:56
  Speaker 1: into a standard intelligence report, the kind that is trusted
• 12:01
  Speaker 1: by military officials, and disseminated it. U.S. Special Operations Command
• 12:07
  Speaker 1: Pacific and the Pentagon did not respond to a request
• 12:11
  Speaker 1: for comment. The rationale for the rapid adoption of AI
• 12:15
  Speaker 1: is that it can help the military make battlefield decisions
• 12:19
  Speaker 1: like which targets to strike or which military assets to
• 12:23
  Speaker 1: move where faster. Officials say the United States cannot afford
• 12:29
  Speaker 1: to fall behind in integrating AI in case it must
• 12:33
  Speaker 1: one day fight China or another adversary who will potentially
• 12:36
  Speaker 1: be able to stay one step ahead of the United States. Now,
• 12:40
  Speaker 1: Those of you who are responsible for the protection of
• 12:43
  Speaker 1: the United States, I'm not talking about your country, your country,
• 12:47
  Speaker 1: your country. I'm talking about the United States and in
• 12:50
  Speaker 1: the states, local and state, county, townships, or whatever. Are
• 12:58
  Speaker 1: you staying on top of all of this for the
• 13:00
  Speaker 1: protection of the citizenry that you're responsible for? I know
• 13:04
  Speaker 1: you are, aren't you? Aren't you? In January, Defense Secretary
• 13:12
  Speaker 1: Reese's agency's artificial intelligence acceleration strategy in a bid to
• 13:18
  Speaker 1: speed up the military's use of AI. We will unleash experimentation,
• 13:23
  Speaker 1: eliminate bureaucratic barriers, focus our investments, and demonstrate the execution
• 13:29
  Speaker 1: approach needed to ensure we lead in military AI. Have
• 13:36
  Speaker 1: you listened today? as it relates to AI and potential
• 13:46
  Speaker 1: war that may have occurred or conflict or whatever the
• 13:49
  Speaker 1: case may be. Well, have you listened? Are you sure?
• 13:53
  Speaker 1: Have you done your research on AI? Well, we're going
• 13:57
  Speaker 1: to talk about why the AI panicked the next time
• 14:02
  Speaker 1: that's going on. You know, it's a good question. But
• 14:07
  Speaker 1: we're going to lay it out for those of you
• 14:09
  Speaker 1: who are of faith. So you're not running around pulling
• 14:12
  Speaker 1: your hair out all afraid. For what? Don't you remember
• 14:17
  Speaker 1: who's in control? I know you do. You do, don't you?