When we get a new tool like the Bridgeport or ShopBot, initially nobody is authorized, or trained as an authorizer, and nobody can authorize others. To solve the chicken and egg problem, an ad hoc group forms around the tool that becomes the initial group of authorized uses. They figure the tool out, determine what is dangerous about it, address safety issues, acquire initial tooling, perform setup and calibration, etc. They clear the way for the membership to use the tool. They also form the pool of initial authorizers to get other people trained and authorized.
"Danger Committee" is the unofficial-official term for the first group of hackers to figure out what might be dangerous about a new tool. The name is tongue-in-cheek because it's a "safety committee" that has been hackerspaceified. What a danger committee is or does will adapt heavily to tool or equipment being researched. There does not have to be extraordinary danger to form a danger committee, it's a response to a need for an authorization process.
As many tools at PS:One allow for any authorized member to authorize any other member, there is not always a danger committee. The need for an authorization process, and what it might be, is a socially determined thing; the rule of thump for PS1 is that a tool as complicated/dangerous as a hot glue gun or a soldering iron, or a screwdriver, does not require authorization, but a 3D printer might, and a table saw certainly does. Authorization might be designed to protect the user, bystanders, the tool itself, or the space/environment.
A danger committee's role is over, mostly, once procedures for authorization and operation have been determined.
Who forms a danger committee? JFDI, or the owner of the tool, or the Board, or the Area Host.
How is a danger committee formal? Once one has been established and starts working, it holds some mandate based on it's instantiation of consensus.
How is a danger committee informal? There doesn't need to be a rule about it in the bylaws, there doesn't need to be a board vote, there doesn't need to be a member vote. It's an example of a social norm that works well without too much documentation.
Who can join a danger committee? Generally, anyone could. Whoever joins should be able to work well with others, consider the point of the task at hand (to get things done, but to protect people, things, the space/environment) and not be there to be a jerk or an obstacle. Genuine research or expertise should be encouraged and embraced, as well as legitimate best practices gleaned from elsewhere.