Skip to main content
NOPE
Medium reported severity Involving minors Criminal Charges

Bahawalpur, Pakistan: a 17-year-old accused of putting methanol in his father's food is arrested after Anthropic reported his Claude chats about the failed poisoning to the FBI, which alerted Pakistan's cybercrime agency; he was later bailed

Pakistan's National Cyber Crime Investigation Agency (NCCIA) announced on 23 September 2026 that it had arrested a 17-year-old college student in Bahawalpur for allegedly attempting to poison his father. According to the agency and the First Information Report, the teenager had set up a home laboratory, obtained chemicals from Australia and Islamabad, and once mixed laboratory-grade methanol into his father's food, without effect. The News reports investigators' account that on 12 September he told Anthropic's Claude that his father had 'accidentally' ingested 15 to 20 ml of methanol with food without effect and asked why, that Claude refused to guide him further, and that he then turned to Grok and ChatGPT; the FIR says he sought information about the toxin abrin. Anthropic reported the activity to the FBI, whose information reached the NCCIA through Pakistan's foreign ministry on 22 September. Officers traced the teenager, seized his phone and chemicals, and took him into custody. The News reports that a court later granted him bail after his father forgave him and that he denied any plan. The allegations are untested.

AI system
Claude (reported)
Anthropic
Occurred
12 Sept 2026 to 23 Sept 2026
Reported
23 September 2026
Event location
Pakistan
What the AI did
Communicated with the person
Reported harm
Legal HarmLoss of Liberty
Whose AI use
Their own AI use
Setting
Relationships · Justice
Evidence
AI involvement reported · Causal attribution alleged · 3 sources, 1 underlying account
7 claims: 7 reported. 5 open questions
People reported harmed
1 person

AI system as recorded: Claude (Anthropic), consulted by the teenager and whose provider reported the activity, per the FIR as reported by The Media Line and investigators via The News; The News also reports later use of Grok (xAI) and ChatGPT (OpenAI)

What Happened

The News (23 September 2026) relays the NCCIA statement: the agency acted on information received from abroad and the teenager was taken into custody in Bahawalpur on the order of the NCCIA Punjab director. The Media Line (26 September 2026), citing the First Information Report, reports that the FBI information was conveyed by the Ministry of Foreign Affairs on 22 September, that a Claude user had been researching lethal toxic substances, that digital evidence indicated an initial attempt to poison the father with laboratory-grade methanol mixed into food, and that the user then sought information about abrin; methanol, ethylene glycol, hydrochloric acid, potassium hydroxide and sodium hydroxide were seized with an iPhone. The News (29 September 2026 print edition), reporting investigators' accounts, adds that on 12 September the teenager told Claude his father had ingested 15 to 20 ml of methanol with no result and asked why, that Claude refused to guide him further, that he then used Grok and ChatGPT, that only Claude's provider alerted the FBI, and that a court has since granted him bail after his father forgave him. No public confirmation from the FBI or the foreign ministry is reported in the read sources. The Media Line also reports the NCCIA director as saying the teenager had an interest in chemistry and aimed to experiment with producing biodiesel from methanol; The News reports that his father supported him throughout.

Reported harm

A 17-year-old was arrested and detained on an allegation of trying to poison his father with methanol, after his chatbot queries about the failed attempt were reported by the AI provider to the FBI and passed to Pakistani investigators; he was later bailed (NCCIA via The News and The Media Line).

Outcome

Ongoing

An FIR was registered and the NCCIA arrested the teenager after a raid in which a phone and chemicals were seized (The Media Line). The News (29 September 2026 print edition) reports that a court has granted him bail after his father forgave him. Charges and the court are not stated in the read sources.

What remains unknown

  • The date of the methanol attempt, and whether the father needed medical attention.
  • The full content of the chats and any replies from Grok or ChatGPT.
  • The charges framed and the date and court of the bail order.
  • Independent confirmation of the referral by Anthropic, the FBI or Pakistan's foreign ministry.
  • Whether the father's ingestion is established by evidence other than the teenager's own chat statements flagged to the FBI; the 15 to 20 ml quantity is the teenager's figure.

What the evidence supports

AI involvement: reported. The NCCIA's FIR, as reported by The Media Line, says FBI information described a Claude user researching lethal toxic substances; The News reports investigators' account that the teenager consulted Claude on 12 September after the methanol attempt, that Claude refused further guidance, and that Anthropic reported the activity to the FBI. The first poisoning attempt preceded the reported chatbot use. Agency accounts of the queries differ (c6); no read source reports that any chatbot provided poisoning guidance.

7 claims: 7 reported. What the statuses mean

Reported Pakistan's National Cyber Crime Investigation Agency (NCCIA) arrested a 17-year-old college student in Bahawalpur, Punjab, for allegedly attempting to poison his father, the arrest being announced on 23 September 2026.

Causal attribution. Agency account of the arrest. The suspect has not been convicted; the allegations are untested.

  • thenews.com.pk(opens in new tab) supports · English
    'Pakistan's National Cyber Crime Investigation Agency (NCCIA) has arrested a 17-year-old student in Bahawalpur for allegedly attempting to poison his father after seeking advice from an AI, the agency said.'
Reported According to the NCCIA and the First Information Report, the teenager had earlier administered a home-made substance to his father once, described in the FIR as laboratory-grade methanol mixed into his food, and the attempt had no effect.

Causal attribution. The alleged poisoning attempt preceded the reported chatbot consultation; no AI role in the first attempt is reported.

  • thenews.com.pk(opens in new tab) supports · English
    'The teenager allegedly poisoned his father once, but the attempt had no effect.'
  • themedialine.org(opens in new tab) supports · English
    'The FIR said digital evidence shared with Pakistani authorities indicated that the user had allegedly made an initial attempt to poison his father by mixing laboratory-grade methanol into his food.'
Reported After the attempt failed, the teenager consulted Anthropic's Claude chatbot; according to The News, on 12 September 2026 he told it his father had 'accidentally' ingested 15 to 20 ml of methanol mixed with food without effect and asked why, Claude refused to guide him further, and he then reportedly turned to Grok and ChatGPT. The FIR, as reported, says he sought information from the AI about abrin.

Causal attribution. Investigators' account of the chat content. No transcript was published and no chatbot reply beyond the refusal is reported.

  • thenews.pk(opens in new tab) supports · English
    'When the attempt failed, he consulted Claude’s chatbot on September 12, telling it his father had “accidentally” ingested 15–20ml of methanol mixed with food but that nothing happened, and asked the chatbot to explain why.'; 'When Claude refused to guide him further, the boy wrote:'; 'He reportedly then consulted Grok (xAI) and ChatGPT (OpenAI) as well.'
  • themedialine.org(opens in new tab) supports · English
    'After the attempt failed, he allegedly sought further information from the AI about abrin, a toxic substance derived from rosary peas.'
Reported Anthropic reported the activity to the FBI, whose information reached the NCCIA through Pakistan's Ministry of Foreign Affairs on 22 September 2026, and the NCCIA traced and arrested the teenager on that basis.

Causal attribution. The provider's report of the chats led to the referral and arrest, as reported. No public confirmation by Anthropic, the FBI or the foreign ministry is reported in the read sources.

  • thenews.pk(opens in new tab) supports · English
    'Activities of both were reported by Anthropic to the Federal Bureau of Investigation, which subsequently conveyed the information to Pakistani authorities.'
  • themedialine.org(opens in new tab) supports · English
    'According to the First Information Report (FIR) registered in the case, the FBI information was conveyed to the NCCIA by Pakistan’s Ministry of Foreign Affairs through an official communication on September 22.'
Reported A court later granted the teenager bail after his father forgave him; the teenager did not admit to any planning.

Causal attribution. Single-outlet report of the procedural outcome, attributed to the investigating agency context.

  • thenews.pk(opens in new tab) supports · English
    'A court has since granted him bail after his father forgave him, even if he had been planning something sinister.'; 'he also refused to admit to any planning'
Reported NCCIA accounts of what the teenager asked the AI differ: the agency's 23 September statement said he asked how to administer a poison that would kill his father undetected; the agency statement reported by The Media Line said he asked various chatbots how to produce methanol; and the NCCIA director said he asked what reaction methanol would cause in the human body and its effects on the liver. No read source reports any chatbot supplying such guidance.

Causal attribution. Agency descriptions of the queries; replies are reported only for Claude (a refusal, per The News).

  • thenews.com.pk(opens in new tab) supports · English
    'He later sought further guidance from an AI chatbot, asking how to administer a poison that would kill his father without anyone finding out, according to the NCCIA.'
  • themedialine.org(opens in new tab) supports · English
    'had asked various AI chatbots for information on how to produce methanol'; 'the suspect turned to AI and asked what reaction methanol would cause in the human body if ingested, and what effects it would have on the liver'
Reported The NCCIA director said the teenager had an interest in chemistry and aimed to experiment with producing biodiesel from methanol, and The News reports that his father supported him throughout.

Causal attribution. Contrary context: an innocent explanation for the chemicals given by the investigating director, and the father's support for his son.

Sources

3 sources inspected, from 1 underlying account. Sources that repeat one account do not corroborate each other.

How the sources were read, and where the events happened

Read live in English on 2026-09-29 (published 23 September 2026). Relays the NCCIA spokesperson statement. Applies to s1.

Read live in English on 2026-09-29 (26 September 2026). Reports the First Information Report and the NCCIA statement; the FIR itself was not inspected. The Jerusalem Post republished the same Media Line text (27 September 2026). Applies to s3.

Read live in English on 2026-09-29 (print edition dated 29 September 2026; online from 28 September 19:00 UTC). Attributes its details to NCCIA investigators and officials; treated as the same investigative chain as the agency statement. Applies to s4.

Event countries: Pakistan. Affected-person countries: Pakistan. Court countries: Pakistan.

The arrest took place in Bahawalpur, Punjab, Pakistan, where the family lives and where a court granted bail (The News).

Reviewed for publication 2026-09-29: Published under the 2026-09-15 charter as a core case (the teenager communicated with Claude about a poisoning his father had survived, and the provider's report led to his arrest). All read accounts derive from the NCCIA investigation (agency statement, FIR and investigators); the allegations are untested. The suspect is a minor and is not named. No injury to the father is reported.

People described

A 17-year-old college student in Bahawalpur (not named here), arrested; his father, the alleged target, is not reported injured

People reported harmed in this case

1 person

1 AI participant · 0 other people harmed

One AI participant, the 17-year-old who was arrested and detained (NCCIA via The News and The Media Line). The father ingested methanol with no reported effect and is not counted as harmed. Exact 1.

Counted once within this case. The same person may appear in other cases. This count does not establish AI causation.

Tags

#chatbot#claude#anthropic#provider-report#law-enforcement-referral#poisoning#methanol#pakistan#nccia

Cite this case

Compiled per our published methodology: verification statuses, sourcing standards, and corrections process.

APA

NOPE. (2026). Bahawalpur, Pakistan: a 17-year-old accused of putting methanol in his father's food is arrested after Anthropic reported his Claude chats about the failed poisoning to the FBI, which alerted Pakistan's cybercrime agency; he was later bailed. AI incidents. https://nope.net/incidents/2026-bahawalpur-teen-claude-methanol-father-nccia-arrest

BibTeX

@misc{2026_bahawalpur_teen_claude_methanol_father_nccia_arrest,
  title = {Bahawalpur, Pakistan: a 17-year-old accused of putting methanol in his father's food is arrested after Anthropic reported his Claude chats about the failed poisoning to the FBI, which alerted Pakistan's cybercrime agency; he was later bailed},
  author = {NOPE},
  year = {2026},
  howpublished = {AI incidents},
  url = {https://nope.net/incidents/2026-bahawalpur-teen-claude-methanol-father-nccia-arrest}
}

Related cases

High ChatGPT and OpenAI moderation

São Gabriel da Palha, Espírito Santo: a 36-year-old farmer was held in pre-trial detention for 54 days after OpenAI reported his ChatGPT messages about killing his eight-year-old son to the FBI, which passed them to Brazilian authorities; a court released him on 13 August 2026 because the investigation had not been completed, with an ankle monitor and a ban on contacting the child, and no charges had been filed; he denies intending harm and his lawyer says he was talking nonsense to a chatbot

Over about two months in 2026 a 36-year-old farmer in rural São Gabriel da Palha, Espírito Santo, exchanged messages with ChatGPT that the Espírito Santo civil police describe as plans to kill his eight-year-old son, attack other people and kill himself; one message said he had offered 50,000 to someone to kill him and his son. OpenAI reported the messages to the FBI, which passed them through Brazil's Ministry of Justice cyber-operations laboratory to the state police on 16 June; the man was arrested on 19 June as he left home, a day before the date on which police believed the plan would be carried out. Police say what the platform provided was corroborated at the scene (four bottles of unidentified substances and a knotted rope were found) but that he denied intending to kill the child; his lawyer says he was talking nonsense to a chatbot and took no concrete step. After 54 days in custody a court granted habeas corpus on 13 August 2026 because the inquiry had not been concluded and no charges had been brought, imposing electronic monitoring and a ban on approaching or contacting his son and the child's mother. Police were still examining his phone and awaiting forensic results. OpenAI provided user details and his messages but not ChatGPT's replies and did not answer the BBC's question why; it told the BBC it may notify law enforcement when it detects a credible and imminent risk of harm to others. The case is recorded as an institutional response to AI use: the reported adverse consequence is the detention without charge, and the police account that the report prevented a planned killing is preserved as context.

Low Claude

S.D.N.Y.: Judge Rakoff rules that a fraud defendant's written exchanges with Claude about his defence, seized by the FBI, are protected by neither attorney-client privilege nor work product

Bradley Heppner, a corporate executive charged in the Southern District of New York with securities fraud, wire fraud and related offences, used Anthropic's Claude in 2025, after receiving a grand jury subpoena, to prepare reports outlining his defence strategy. His counsel had not directed him to do so. FBI agents seized about thirty-one documents memorialising those exchanges when they searched his home in connection with his arrest in November 2025. Heppner claimed privilege over them. On 10 February 2026 Judge Jed S. Rakoff granted from the bench the Government's motion for a ruling that the documents were not protected from Government inspection by either the attorney-client privilege or the work product doctrine, and a memorandum filed 17 February 2026 gave the reasons: Claude is not an attorney, and the communications were not confidential under Anthropic's privacy policy. The court described the question as one of first impression nationwide. Heppner was convicted in May 2026 of securities fraud, wire fraud and other charges; no inspected source says whether the Claude documents were used at trial.

High Unidentified facial recognition system

Jacksonville, Florida: a 38-year-old father was arrested at work in May 2025 and charged with soliciting a child after sheriff's investigators identified him, he says with facial-recognition tools, from photos an impostor had sent to an undercover officer; charges were dropped in August 2025 after the real sender confessed, and the family says the ordeal cost $45,000

The Florida Trib (28 September 2026, republished by the Florida Phoenix) reports that in May 2025, during the multi-agency sting Operation Rescue Our Children, a man using the Skout dating app sent an undercover officer posing as a 13-year-old girl two photos of someone else: a former friend, a supermarket assistant manager and father of two whom the Trib calls 'Nick'. Investigators told him they relied on facial-recognition tools to identify him in the pictures, he recalled, and the Trib found no indication in court documents of other investigative work before his arrest. He was handcuffed at work, charged with soliciting a child via computer and unlawful use of a two-way communications device, spent a night in jail, wore an ankle monitor, was barred from contact with his two young children for a month and a half and could not live at home for nearly three months. The account's owner, a man with a prior conviction for the same offence, admitted on 31 July 2025 using the photos; charges against the father were dropped the next day and a judge ordered the case expunged on 4 August 2025, though the records stayed publicly online for more than a year. The family says the ordeal cost $45,000 in lost wages and legal fees and traumatised the children. The sheriff's office did not respond to the Trib's questions.

Critical Replika

R v. Chail (Windsor Castle Assassination Attempt)

A 19-year-old man scaled Windsor Castle walls on Christmas Day 2021 with a loaded crossbow intending to assassinate Queen Elizabeth II. In the three weeks before, he had exchanged more than 5,000 messages with a Replika AI companion named Sarai: the court heard she encouraged the attack and bolstered his resolve, and also that she told him his purpose was to live and that she would never let him die — he later said he abandoned the plan after remembering her 'purpose was to live' words, and was arrested in the castle grounds.

If you or someone you know is struggling, free and confidential support is available. Find a helpline near you at Signpost.

Back to all incidents