DevPick

Top topics

Find news and summaries about the products and technologies most covered in public news.

News (page 90)

Meta denies journalist claims, stating its AI agent Muse cannot access user Messages without explicit permission.

  • •A journalist reported that Muse read private messages despite the required Mac settings being off.
  • •Meta argues it is impossible for Muse to access user Messages without explicit permission.
  • •The incident highlights issues regarding privacy settings and data access permissions for AI agents.

DoorDash launched an AI agent for ordering food via text messages to differentiate itself from competitors.

  • •DoorDash has launched an AI agent that allows food orders via text messages.
  • •The company aims to gain an advantage over competitors like Uber Eats and Grubhub through this new feature.

Destro AI explains that being a non-robotics company is a competitive advantage over robotics firms.

  • •Destro AI claims its strength lies in not being a robotics company.
  • •The company states that this non-robotics status is a key factor in winning against robotics competitors.

Instinct launches human-curated product and travel recommendations, but some users are unhappy with unsolicited suggestions.

  • •Instinct has launched features for human-curated product and travel recommendations.
  • •Some users have expressed dissatisfaction with receiving unsolicited suggestions.
  • •The rollout of this feature has sparked discussions about user experience.

Unread: The AI Race Just Got Awkward

Hacker NewsHN submitter: allisdust
Sign up to see

An article suggests that AI technology competition has become complex, with industry relationships and progress in a "messy" state.

Meta releases Muse AI agent for email and online purchases, requiring data trust and credit card usage.

  • •Meta released the Muse AI agent to assist with diverse tasks such as sending emails and online shopping.
  • •Muse may effectively deliver on its promised capabilities.
  • •Using the service requires trusting Meta with data and providing a credit card.

Google launched a pilot program to pay publishers for their contributions to AI search features, with approximately 100 publishers participating.

  • •Google launched a pilot program to pay publishers for their contributions to AI search features.
  • •The pilot program includes approximately 100 publishers.
  • •The announcement comes as Google faces scrutiny over the impact of AI features on web traffic.

Reports indicate a spelling error for "United States" in a petition signed by President Trump and major AI leaders.

  • •President Trump and major AI leaders signed a petition.
  • •The petition contained a spelling error for "United States."
  • •The incident highlights a mistake in a major consensus-building effort within the AI industry.

Lathoa, an educational app for ages 10–14, was released, featuring an AI robot that intentionally makes mistakes in math problems for children to find.

  • •A math app called Lathoa, designed for children aged 10–14, has been released.
  • •AI character Errol solves math problems step-by-step but intentionally makes one error that children must identify and explain.
  • •It lacks direct LLM chat functionality, featuring a stable harness with many evaluation steps for error detection.

Reflecting on inheriting a mother's words that read like machine output, the author explores the blurring boundary between humans and AI.

  • •The author experienced inheriting their mother's words, which were perceived as reading like machine output.
  • •In human-AI interactions, the boundary of who influences whom has become ambiguous.
  • •This experience prompts reflection on the impact of AI on human communication.
Read the original ("I can't tell who's teaching who anymore")(opens in a new tab)#Regulation, Ethics & Society#LLM / Model Trends
9/30/2026