Could Facebook have been aware of frightening direct-message threats made by a shooter accused of killing 19 children and two teachers at an elementary school in Texas? Could it have forewarned authorities?
Texas Gov. Greg Abbott released the online remarks delivered minutes before the Wednesday assault, referring to them as postings, which are routinely sent to a large audience. Facebook intervened, noting that the gunman sent one-on-one direct messages rather than public posts and that they were found "after the tragic incident."
The most recent mass shootings in the United States by active social-media users may increase pressure on social media companies to tighten their scrutiny of online communications, even as conservative politicians, including Abbott, are urging social platforms to relax their restrictions on some speech.
Meta, the parent company of Facebook, has stated that it checks people's private conversations for potentially hazardous information such as links to malware or photographs of child sexual exploitation. However, cloned photos may be recognized using unique identifiers a type of digital signature making them very simple for computer systems to identify. Attempting to understand a series of frightening phrases that may be a joke, satire, or song lyrics is a significantly more challenging challenge for artificial intelligence systems.
Facebook, for example, might flag things like "going to kill" or "going to shoot," but without context, something AI in general struggles with there would be far too many false positives for the firm to analyze.
As a result, Facebook and other platforms rely on user reporting to identify threats, harassment, and other breaches of the law or their policies. As the recent shootings demonstrate, that frequently comes too late, if at all.
Planned Obsolescence
Even this type of surveillance may become useless shortly, as Meta wants to implement end-to-end encryption on its Facebook and Instagram messaging systems next year. Because of this encryption, no one can decipher people's messages save the sender and the recipient - not even Meta. WhatsApp, which is also owned by Meta, already includes this type of encryption.
A recent Meta-commissioned paper emphasized the benefits of such privacy while also highlighting certain hazards, such as users abusing the encryption to sexually exploit children, support human trafficking, and propagate hate speech.
Apple's iMessage system has long maintained end-to-end encryption. As a result, the iPhone maker is at odds with the Justice Department over texting privacy. Following the fatal shooting of three US sailors at a Navy facility in December 2019, the Justice Department insisted on access to data from two locked and encrypted iPhones belonging to the suspected gunman, a Saudi aviation student.
Security experts believe this is possible if Apple creates a "backdoor" that allows access to texts made by accused criminals. With a court order and a secret key, they could decrypt encrypted data.
However, the same experts warned that incorporating backdoors into encryption methods renders them fundamentally unsafe. The very existence of a backdoor draws the attention of the world's spies and criminals to the mathematical keys that may unlock it. When they do, everyone's information becomes available to anybody who has a secret key.
